Detailed Description
In order to make the present application solution better understood by those skilled in the art, the following description will be made in detail and with reference to the accompanying drawings in the embodiments of the present application, it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, shall fall within the scope of the present application.
It should be noted that the terms "first," "second," and the like in the description and claims of the present application and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that embodiments of the present application described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The technical scheme provided by the application can be realized by adopting a large model technology, wherein a large model refers to a deep learning model with large-scale model parameters, and the deep learning model can generally contain hundreds of millions, billions, trillions and even billions of model parameters. The large model can be called as a basic stone model/basic model (Foundat ion Model), the large model is pre-trained through large-scale non-labeled corpus, a pre-trained model with more than one hundred million parameters is produced, the model can adapt to a wide downstream task, and the model has good generalization capability, such as a large-scale language model (Large Language Model, abbreviated as LLM), a multi-modal pre-training model (mult i-model) and the like.
It should be noted that, when the large model is actually applied, the pretrained model can be finely tuned by a small number of samples, so that the large model can be applied to different tasks. For example, the large model can be widely applied to the fields of natural language processing (Natural Language Process ing, abbreviated as NLP), computer vision, voice processing and the like, and can be particularly applied to the tasks of the fields of computer vision such as vision question-answering (Visual Quest ion Answering, abbreviated as VQA), image description (IC), image generation and the like, and can also be widely applied to the tasks of the fields of natural language processing such as emotion classification based on texts, text abstract generation, machine translation and the like. Thus, major application scenarios for large models include, but are not limited to, digital assistants, intelligent robots, searches, online education, office software, electronic commerce, intelligent design, and the like. In the embodiment of the application, explanation will be given by taking data processing through a machine learning model in a dialogue scene as an example.
First, partial terms or terminology appearing in describing embodiments of the present application are applicable to the following explanation:
the ruler code list recommendation is used for carrying out matching recommendation according to the stature files of the user, such as height, weight, chest circumference and other information and the product size list;
the historical purchasing sequence of the user can be a product stock unit sequence of historical purchasing of the user on the platform;
stock keeping units (Stock Keeping Unit, abbreviated SKUs), which can be unique codes that distinguish different products or products, can be combinations of numbers, letters or symbols, each SKU corresponds to a person-specific product, helping merchants manage inventory, sell and track products;
a target-attention framework (target-attention) may be applied in the recommendation task, which may be a target attention operation for candidate products (items) and click sequence products of the user.
Example 1
According to one method of the embodiment of the application, a product recommendation method is provided. As an alternative embodiment, the recommendation method of the product may include, but is not limited to, application to the application scenario shown in fig. 1. Fig. 1 is a schematic diagram of an application scenario of a product recommendation method according to an embodiment of the present application, as shown in fig. 1, in the application scenario, a terminal device 102 may, but is not limited to, communicate with a server 106 through a network 104, for example, may be used to transmit object feature information, a product to be recommended, and the like, and the server 106 may, but is not limited to, perform an operation on a database 108, for example, a data writing operation or a data reading operation. The terminal device 102 may include, but is not limited to, a man-machine interaction screen, a processor, and a memory. The man-machine interaction screen may be used for displaying a candidate product set and a product to be recommended on the terminal device 102, but is not limited to the man-machine interaction screen. The processor may include, but is not limited to, a processor configured to perform a corresponding operation in response to the man-machine interaction operation, or generate a corresponding instruction and send the generated instruction to the server 106. The memory is used for storing related processing data such as a recommendation model and the like.
As an alternative, the following steps in the product recommendation method may be performed on the server 106: step S102, determining a recommended object which needs to execute a recommending operation; step S104, object characteristic information of a recommended object is called; step S106, analyzing object feature information by using a recommendation model, and determining a candidate product set and recommendation information of different candidate products in the candidate product set; step S108, determining at least one product to be recommended from the candidate product set based on the recommendation information; step S110, recommending the product to be recommended to the recommended object.
By adopting the mode, determining a recommended object which needs to execute the recommended operation; retrieving object feature information of a recommended object, wherein the object feature information is used for at least representing preference degrees of the recommended object on different types of products; analyzing object feature information by using a recommendation model, determining a candidate product set and recommendation information of different candidate products in the candidate product set, wherein the recommendation information is used for predicting the probability that a recommendation object performs transaction operation on the candidate products; determining at least one product to be recommended from the candidate product set based on the recommendation information; and recommending the product to be recommended to the recommended object. That is, in this embodiment, object feature information of a recommended object is determined, and the object feature information is analyzed by using a recommendation model to determine recommendation information corresponding to different candidate products, and a target candidate product is determined based on the recommendation information, so that a technical effect of improving accuracy of product recommendation is achieved, and a technical problem of low accuracy of product recommendation is solved.
In another alternative embodiment, FIG. 2 illustrates in a block diagram one embodiment of a computing node in a computing environment 201 using a computer terminal (or mobile device). Fig. 2 is a block diagram of a computing environment, as shown in fig. 2, where the computing environment 201 includes a plurality of computing nodes (e.g., servers) running on a distributed network (shown as 210-1, 210-2, …) in accordance with an embodiment of the present application. The computing nodes each contain local processing and memory resources and end user 202 may run applications or store data remotely in computing environment 201. The application may be provided as a plurality of services 220-1, 220-2, 220-3, and 220-4 in computing environment 201, representing services "A", "D", "E", and "H", respectively.
End user 202 may provide and access services through a web browser or other software application on a client, in some embodiments, provisioning and/or requests of end user 202 may be provided to portal gateway 230. Ingress gateway 230 may include a corresponding agent to handle provisioning and/or request for services (one or more services provided in computing environment 201).
Services are provided or deployed in accordance with various virtualization techniques supported by the computing environment 201. In some embodiments, services may be provided according to Virtual Machine (VM) based virtualization, container based virtualization, and/or the like. Virtual machine-based virtualization may be the emulation of a real computer by initializing a virtual machine, executing programs and applications without directly touching any real hardware resources. While the virtual machine virtualizes the machine, according to container-based virtualization, a container may be started to virtualize the entire operating system (Operat ing System, abbreviated as OS) so that multiple workloads may run on a single operating system instance.
In one embodiment based on container virtualization, several containers of a service may be assembled into one computing unit (e.g., kubernetes Pod). For example, as shown in FIG. 2, service 220-2 may be equipped with one or more computing units (Pod) Pod240-1,240-2, …,240-N (collectively Pod). The Pod may include an agent 245 and one or more containers 242-1,242-2, …,242-M (collectively referred to as containers). One or more containers in the Pod handle requests related to one or more corresponding functions of the service, and the agent 245 generally controls network functions related to the service, such as routing, load balancing, etc. Other services may also be equipped with Pod similar to Pod.
In operation, executing a user request from end user 202 may require invoking one or more services in computing environment 201, and executing one or more functions of one service may require invoking one or more functions of another service. As shown in FIG. 2, service "A"220-1 receives a user request of end user 202 from ingress gateway 230, service "A"220-1 may invoke service "D"220-2, and service "D"220-2 may request service "E"220-3 to perform one or more functions.
The computing environment may be a cloud computing environment, and the allocation of resources is managed by a cloud service provider, allowing the development of functions without considering the implementation, adjustment or expansion of the server. The computing environment allows developers to execute code that responds to events without building or maintaining a complex infrastructure. Instead of expanding a single hardware device to handle the potential load, the service may be partitioned to a set of functions that can be automatically scaled independently.
It should be noted that, the user information (including, but not limited to, user equipment information, user personal information, etc.) and the data (including, but not limited to, data for analysis, stored data, displayed data, etc.), such as weather forecast results, are information and data authorized by the user or fully authorized by the parties, and the collection, use and processing of the relevant data need to comply with relevant laws and regulations and standards of relevant countries and regions, and is provided with corresponding operation entries for the user to select authorization or rejection.
Under the above operating environment, the present application provides a product recommendation method. Fig. 3 is a flow chart of a method of recommending a product according to an embodiment of the present application. As shown in fig. 3, the method may include the steps of:
In step S302, a recommendation object for which a recommendation operation needs to be performed is determined.
In the technical solution provided in step S302 in the present application, a recommendation object that needs to execute a recommendation operation is determined. The recommending operation may refer to recommending a product to a recommending object, and may be a product recommending operation, for example, may be a clothing recommending operation, an electronic article recommending operation, or the like. The recommended object may be an object for performing a recommendation operation, for example, may be a consumer in an e-commerce platform, a user of a streaming media platform, etc., and is merely illustrated herein, and the type of the recommendation operation, and the identity of the recommended object are not specifically limited.
Alternatively, when a recommendation operation is to be performed, a recommendation object that needs to perform the recommendation operation is determined, for example, when a product is required to be recommended for a consumer in the e-commerce platform, identity information of a mobile terminal currently logged in to the e-commerce platform may be determined to determine the consumer that needs to perform the product recommendation operation, or in the music platform, an audio listener that needs to perform the recommendation operation may be determined to need to perform the recommendation operation.
In step S304, object feature information of the recommended object is called, where the object feature information is used to at least characterize the preference degree of the recommended object for different types of products.
In the technical solution provided in step S304, after determining the recommended object, the object feature information of the recommended object is called. The object feature information may be multi-dimensional feature information of the recommended object, may be used for at least characterizing the preference degree of the recommended object for different types of products, and may include a user size file, a user portrait, a user history purchase sequence, etc. of the recommended object, which is merely illustrative of the type of the object feature information, and is not particularly limited. The preference level may be used to characterize at least the purchase preferences of the recommended object, and may include wear style preferences, product brand preferences, etc., without specific limitation to the meaning of the characterization of the preference level. Different types of products may be different brands, different styles, different properties.
Alternatively, object feature information of the recommended object may be determined and stored by feature extraction or the like in advance. When a recommended object for which a recommendation operation is required to be performed is determined, object feature information of the recommended object may be determined from a pre-stored location. The determination manner of determining the object feature information is not particularly limited here.
For example, using feature engineering, object feature information of a recommended object may be determined and stored in a database. When the recommended object for which the recommending operation needs to be performed is determined, object characteristic information of the recommended object can be called from the database.
In step S306, the recommendation model is used to analyze the object feature information, determine a candidate product set, and recommendation information of different candidate products in the candidate product set, where the recommendation information is used to predict the probability that the recommendation object performs a transaction operation on the candidate products.
In the technical solution provided in step S306 of the present application, after the object feature information of the recommended object is called, the product to be recommended to the recommended object may be determined by using the recommendation model. The object feature information may be analyzed using a recommendation model to determine a set of candidate products, as well as recommendation information for different candidate products in the set of candidate products. The recommendation model (Deep Interest Network, abbreviated as DIN) may be obtained by training any neural network model such as a Target-attention frame (Target-attention), or a deep learning network (deep), a feature cross model (Workflow Descript ion Language, abbreviated as WDL), and the like, and may be used to determine a candidate product set recommended to a recommendation object, and recommendation information of different candidate products, where the structure of the recommendation model is not specifically limited. The candidate product set may be a list of candidate products for performing a recommendation operation, may be selected from the entire product library by a recommendation model, and may be selected as a potential recommendation. The recommendation information may be used to determine the probability of the recommended object performing a transaction operation on the candidate product, for example, text information may be recommended, or 1 may represent recommendation, 0 may represent non-recommendation, 2 may represent highly recommended digital recommendation information, etc., which is only for illustration and not specific limitation on the expression form of the recommendation information. The candidate products may be represented by product stock keeping units (simply referred to as product SKUs). The transaction operation may be a purchase operation.
Optionally, in the e-commerce platform, the recommendation model may analyze information such as personal data, historical purchase records, browsing behaviors, and the like of the user in the object feature information, so as to select a candidate product set from the product library, and determine recommendation information of different candidate products in the candidate product set.
For example, for a recommended object that likes fitness, object feature information of the recommended object may be analyzed using a recommendation model to screen products related to fitness equipment, health food, and the like from a product library as candidate products, to obtain a candidate product set. After the candidate product set is determined, the recommendation model determines recommendation information of different candidate products in the candidate product set.
Optionally, the target attention frame may be learned in an incremental learning manner to obtain the recommendation model. It should be noted that, in addition to the incremental learning manner, the target attention frame may be retrained every day to obtain the recommended model, and the model training manner is not particularly limited here.
Step S308, determining at least one product to be recommended from the candidate product set based on the recommendation information.
In the technical scheme provided in the step S308, recommendation information of different candidate products in the candidate product set is determined, and at least one product to be recommended is determined from the candidate product set based on the recommendation information.
Optionally, the preference degree of different types of products of the recommended object is modeled by introducing an attention mechanism, so as to obtain a recommendation model. And determining a candidate product set based on the preference degree of the recommended object on different types of products by using a recommendation model, and determining recommendation information of different candidate products in the candidate product set, and determining at least one product to be recommended from the candidate product set based on the recommendation information.
As can be seen from the foregoing, when the embodiment executes the recommendation operation on the recommendation object, not only the user portrait of the recommendation object, such as the user statue, but also the object feature information of the recommendation object, such as the preference of the recommendation object, is considered, and the object feature information is analyzed through the recommendation model, so as to obtain an accurate recommendation result, thereby achieving the technical effect of improving the accuracy of product recommendation, and solving the technical problem of low accuracy of product recommendation.
Step S310, recommending the product to be recommended to the recommended object.
In the technical solution provided in step S310, at least one product to be recommended that is interested in the recommended object may be determined from the candidate product set based on the recommendation information, and the product to be recommended may be recommended to the recommended object, for example, the product to be recommended may be displayed in the mobile terminal operated by the recommended object, so as to recommend the product to be recommended to the recommended object.
Because the product size table uploaded by the merchant has the problem of low accuracy, if the product to be recommended is determined only according to the product size table and the stature information of the recommended object, the technical problem of low accuracy of the recommended size of the product exists. That is, the embodiment not only determines the product to be recommended to the recommended object based on the stature information of the recommended object, but also based on the multi-dimensional object characteristic information such as the preference degree of the recommended object to different types of products, thereby achieving the technical effect of improving the accuracy of product recommendation and solving the technical problem of low accuracy of product recommendation.
For example, determining a recommended object that needs to perform a recommendation operation, and retrieving object feature information of the recommended object is: for casual wind clothes, loose style clothes are favored. The method comprises the steps of analyzing object feature information by using a recommendation model, determining a candidate product set of casual wind clothing, determining recommendation information of different candidate products in the candidate product set, determining recommendation information of a candidate product I as general recommendation, candidate information of a candidate product II as very recommendation, candidate information of a candidate product III as non-recommendation, determining products to be recommended as a candidate product I and a candidate product II from the candidate product set based on the candidate information, and recommending the candidate product I and the candidate product II to a recommendation object for selection of the recommendation object.
In the embodiment of the application, determining a recommended object which needs to execute a recommending operation; retrieving object feature information of a recommended object, wherein the object feature information is used for at least representing preference degrees of the recommended object on different types of products; analyzing object feature information by using a recommendation model, determining a candidate product set and recommendation information of different candidate products in the candidate product set, wherein the recommendation information is used for predicting the probability that a recommendation object performs transaction operation on the candidate products; determining at least one product to be recommended from the candidate product set based on the recommendation information; and recommending the product to be recommended to the recommended object. That is, in this embodiment, object feature information of a recommended object is determined, and the object feature information is analyzed by using a recommendation model to determine recommendation information corresponding to different candidate products, and a target candidate product is determined based on the recommendation information, so that a technical effect of improving accuracy of product recommendation is achieved, and a technical problem of low accuracy of product recommendation is solved.
The above-described method of this embodiment is further described below.
As an optional implementation manner, step S306, using the recommendation model to analyze the object feature information, determines recommendation information of different candidate products in the candidate product set, including: acquiring attribute information of different candidate products; analyzing object feature information and attribute information of different candidate products by using a recommendation model, and determining recommendation scores of the different candidate products; determining recommendation information of different candidate products based on the recommendation scores, wherein the recommendation information at least comprises: whether to recommend a probability of a candidate product.
In this embodiment, attribute information of different candidate products is obtained, attribute information and object feature information of the different candidate products may be analyzed using a recommendation model, recommendation scores corresponding to the different candidate products are obtained, and recommendation information of the different candidate products may be determined based on the recommendation scores. Wherein the recommendation information may include at least a probability of whether to recommend the candidate product. The recommendation score may be used to characterize the degree of recommendation for the candidate product, e.g., may be 0, 1, 2, etc., with a higher recommendation score indicating a higher degree of recommendation for the candidate product or a lower recommendation score indicating a lower degree of recommendation for the candidate product. The size of the recommendation score is merely illustrative and is not particularly limited. The attribute information may be a feature of the SKU dimension of the candidate product and may include at least one of: identification information, size attributes, brands, candidate product identity information (Ident ity Document, abbreviated as ID), store information, SKU ID, etc., are given here by way of example only, and the contents of the attribute information are not particularly limited.
Optionally, the object feature information is analyzed by using a recommendation model, a candidate product set is determined, attribute information of different candidate products in the candidate product set is obtained, the attribute information and the object feature information are analyzed by using the recommendation model, so as to determine recommendation scores of the different candidate products, and recommendation information is determined based on the recommendation scores.
Optionally, pre-stored object feature information is obtained according to identity information of the recommended object, pre-stored attribute information of different candidate products is determined according to product SKUs of the candidate products, and the attribute information and the object feature are analyzed by using a recommendation model to score the different candidate products, so that recommendation scores of the SKUs of the products under the recommended object are obtained.
For example, attribute information of the first candidate product, the second candidate product and the third candidate product is determined respectively, the first candidate product, the second candidate product and the third candidate product are scored by combining object feature information of the first recommendation object with a recommendation model, and a recommendation score of 59 for the first candidate product, a recommendation score of 80 for the second candidate product and a recommendation score of 95 for the third candidate product are obtained. It should be noted that, the determination of the recommendation score and the recommendation information is merely illustrative, and is not limited in particular.
As an alternative embodiment, determining recommendation scores for different candidate products using a recommendation model to analyze object feature information and attribute information for the different candidate products includes: analyzing object feature information by using a recommendation model, and determining historical transaction preference of a recommendation object on at least one historical product; analyzing attribute information by using a recommendation model, and determining transaction preferences of a recommendation object on different candidate products; determining a similarity between the historical transaction preferences and the transaction preferences using a recommendation model; and determining recommendation scores of candidate products based on the similarity, wherein the recommendation scores are used for determining recommendation information matched with the similarity, and the similarity and the recommendation information are positively correlated.
In this embodiment, the analysis may be performed by analyzing object feature information and attribute information of different candidate products using a recommendation model by: the method includes the steps of analyzing object feature information by using a recommendation model, determining historical transaction preferences of a recommendation object for at least one historical product, analyzing attribute information by using the recommendation model to determine transaction preferences of the recommendation object for different candidate products, determining similarity between the historical transaction preferences and the transaction preferences by using the recommendation model, determining recommendation scores of the candidate products based on the similarity, and determining recommendation information matched with the similarity based on the recommendation scores. Wherein historical transaction preferences may be used to characterize the type of consumption that a user tends to purchase for such historical products, such as, for example, for athletic wear of brand one, recommended subject tends to purchase loosely. The historical product may be a product that the recommending object purchased at a historical time. The similarity is positively correlated with the recommendation score, e.g., the higher the similarity, the higher the corresponding recommendation score.
Optionally, in addition to comparing the similarity between the historical transaction preference and the transaction preference to determine the recommendation score, the embodiment may also calculate the similarity between the attribute information of the candidate product in different dimensions and the attribute information of the historical product purchased in the history, where the more similar the candidate product is, the higher the recommendation score is, so as to determine the recommendation score, or may combine the two determinations to determine the recommendation scores corresponding to the two determinations, and perform weighted average on the obtained two recommendation scores to obtain the final recommendation score. That is, the embodiment determines the recommendation score through various modes, thereby determining whether to recommend the candidate product to the recommended object, further achieving the technical effect of improving the accuracy of the recommended product, and solving the technical problem of low accuracy of the recommended product.
For example, the recommendation model is used to analyze the object feature information, determine that the historical transaction preference of the recommendation object for the casual clothing of the brand one is a slimming model, the recommendation model is used to analyze the attribute information of the candidate product, determine that the transaction preference of the candidate product of the brand A for the recommendation object is a slimming model, determine that the historical transaction preference of the candidate product one for the historical product is the same, and determine that the recommendation score is 95.
For another example, historical product one, historical product B, and historical product C, which have high relevance to the candidate product style, store style, may be determined first, and object feature information may be analyzed using a recommendation model to determine historical transaction preferences for historical product one, historical product B, and historical product C, where the historical transaction preferences may include size preferences. And analyzing attribute information of the first candidate product, the second candidate product and the third candidate product by using a recommendation model, and determining transaction preference of a recommendation object on the first candidate product, the second candidate product and the third candidate product, wherein the first candidate product, the second candidate product and the third candidate product can be garments in the same style with different sizes. Based on the similarity, recommendation scores of candidate products of different sizes are determined, so that candidate products of which the sizes meet the preference of the recommended object among the plurality of candidate products are determined and recommended to the recommended object.
As an alternative embodiment, analyzing object characteristic information using a recommendation model to determine historical transaction preferences of a recommendation object for at least one historical product, comprising: and analyzing a historical transaction sequence in the object characteristic information by using the recommendation model, and determining the historical transaction preference of the recommendation object on at least one historical product, wherein the historical transaction sequence is used for representing the purchase condition of the recommendation object on the at least one historical product.
In this embodiment, a historical transaction sequence in the object characteristic information is determined, and the historical transaction sequence is analyzed using a recommendation model to determine a historical transaction preference of the recommendation object for at least one historical product. The historical transaction sequence can be a product SKU sequence of a recommending object purchased in the platform in a historical manner and can be used for representing the purchasing condition of the recommending object on at least one historical product.
Alternatively, the recommended object may have different purchase preferences for different types or brands of products, such as a larger or smaller dressing preference, and thus, the historical transaction sequence of the recommended object may be analyzed to determine historical transaction preferences of the recommended object for different historical products, based on which the recommended object is prone to select products from different candidate products.
In order to solve the problem that the accuracy of recommendation is low when recommending products to a recommendation object due to the fact that consumers have different dressing preferences for different types or brands of products, such as clothes, and the use of size table matching cannot meet the personalized requirements of the users, in this embodiment, based on the historical purchasing sequence of the recommendation object, understanding of purchasing habits of the recommendation object is enhanced so as to consider the dressing size preference of the users, and through feature screening, the problem of uncontrollable product size table coverage rate and accuracy is avoided, so that the technical effect of improving the accuracy of product recommendation is achieved, and the technical problem of low accuracy of product recommendation is solved.
As an alternative embodiment, using a recommendation model to analyze historical transaction sequences in object feature information, determining historical transaction preferences of a recommendation object for at least one historical product includes: acquiring a historical browsing sequence with the correlation degree with the candidate product higher than a correlation degree threshold value in the object characteristic information, wherein the historical browsing sequence is used for representing the browsing condition of a recommended object on at least one historical product in a target time period; the historical browsing sequence is analyzed using the recommendation model to determine historical transaction preferences for at least one historical product.
In this embodiment, when analyzing the historical preferences of the recommended object for different historical products, the historical transaction preferences of the recommended object for at least one historical product may be determined not only based on the historical purchase sequence, but also based on the historical browsing sequence of the recommended object.
Optionally, in the target time period, a history browsing record of the recommended object in the sales platform is obtained, and a history browsing product, of which the similarity with the candidate product is higher than a correlation threshold, in the history browsing record is determined, so that a history browsing sequence is constructed. The historical browsing sequence is analyzed using a recommendation model to determine historical transaction preferences of a recommendation object for at least one historical product, e.g., the more browses for the same type of historical product, the more favored the recommendation object may be for purchasing that type of product. The target period may be a set historical period, for example, within one month or one worship, and the size of the target period is not particularly limited here.
In this embodiment, the historical transaction preference of the recommended object for different historical products may be determined based on the historical transaction sequence of the recommended object, the historical transaction preference of the recommended object for different historical products may be determined based on the historical browsing behavior of the recommended object, the historical transaction preference may be determined based on information such as the historical search behavior, etc., where the condition for determining the historical transaction preference is not specifically limited, and the manner of determining the historical transaction preference should be within the protection scope of the present application as long as the manner is based on the historical behavior of the recommended object.
Optionally, the manner of determining the historical transaction preference may be selected according to the candidate product of the recommended object, for example, if the candidate product is an electronic product, focusing on the product performance index, the historical transaction preference may be determined together based on the historical browsing record of the recommended object and the historical transaction sequence; if the candidate products are different sized garments, focusing on the product size, historical transaction preferences of the recommending object for different types of garments may be determined based solely on the recommending object's historical transaction sequence.
For example, a history browsing sequence of the recommended object in the last month is obtained, based on the history browsing sequence, it can be determined that the recommended object has a large number of times of browsing the solid wood furniture, a small number of times of browsing the plastic furniture and a general number of times of browsing the marble furniture recently, then it can be determined that the recommended object has a high historical transaction preference for the solid wood furniture, the marble furniture has a general historical transaction preference, and the plastic furniture has a low historical transaction preference, then the solid wood furniture can be selected from the candidate products and recommended to the recommended object. It should be noted that the historical browsing sequence is only a single evaluation dimension, and the determination of the historical transaction preference may be considered from multiple evaluation dimensions.
As an alternative embodiment, determining historical transaction preferences and transaction preferences using a recommendation model, the similarity between the two, comprises: using a recommendation model to determine a historical product with the same type as the candidate product in at least one historical product as a target historical product; determining a target historical transaction preference of a target historical product; using the recommendation model, a similarity between the target historical transaction preferences and the transaction preferences of the different candidate products is determined.
In this embodiment, in determining the recommendation score based on the similarity, at least one historical product may be filtered to improve the accuracy of the recommendation score determination. The process of screening at least one historical product may include the steps of: and determining the historical products with the same type as the candidate products in at least one historical product by using a recommendation model as target historical products, determining target historical transaction preferences corresponding to the target historical products, and determining the similarity between the target historical transaction preferences and transaction preferences of different candidate products by using the recommendation model.
Optionally, to improve the efficiency of data processing, in determining the historical transaction preference, it may be determined that only the historical transaction preference of the target historical product of the same type as the candidate product is determined, and the product to be recommended in the candidate product is determined by determining the similarity between the historical transaction preference of the target historical product and the transaction preference of a different candidate product. That is, according to the category of the product, the embodiment utilizes the recommendation model to evaluate the similarity between the candidate products of each size and the target historical products in the historical purchasing sequence of the object, implicitly considers the transaction preference of the object, improves the processing efficiency of the data, and achieves the purpose of accurately determining the product to be recommended.
As an alternative embodiment, determining recommendation scores for different candidate products using a recommendation model to analyze object feature information and attribute information for the different candidate products includes: determining the size in the attribute information, the historical purchase sequence in the object feature information and the size information in the object feature information; determining a first similarity between the size and the size information of the candidate product using the recommendation model, and determining a second similarity between the candidate product and the historical product in the historical purchase sequence using the recommendation model; weighting the first similarity and the second similarity to obtain the similarity; based on the similarity, a recommendation score for the candidate product is determined.
In this embodiment, when the recommendation model is used to analyze object feature information and attribute information of different candidate products, the similarity between the historical transaction preference and the transaction preference may be simply compared, or the candidate products may be directly compared with the historical products, and the candidate products may be compared with the historical products to determine recommendation scores of the candidate products through the following steps: the size of the candidate product in the attribute information, the historical purchase sequence in the object feature information, and the size information in the object feature information are determined. And determining a first similarity between the size and the size information of the candidate product and a second similarity between the candidate product and the historical product by using a recommendation model, weighting the first similarity and the second similarity to obtain a final similarity, and determining a recommendation score of the recommended product based on the similarity. The size information can be used for representing the information of the size, width, length and the like of the product.
Alternatively, the first similarity between the candidate product and the size information of the historic product in the historic purchase sequence is determined, respectively, for example, when the candidate product is L-code and the size information of the historic product is also L-code, the first similarity may be determined to be 100%. Meanwhile, a second similarity between the candidate product and the history product may be determined, for example, when the candidate product is sports pants and the history product is sports wear, the second similarity may be determined to be 70%, or when the candidate product is next-to-the-skin sweater and the history product is next-to-the-skin autumn sweater, the second similarity may be determined to be 60%, and the number sizes are merely illustrative and not particularly limited. The first similarity and the second similarity may be weighted to determine a final similarity, thereby improving accuracy in determining the similarity.
In this embodiment, the first similarity and the second similarity are determined, so that the purchase size is recommended based on the stature information of the recommended object, the size of the candidate product and the historical purchase behavior sequence, and the purpose of improving the accuracy of product recommendation is achieved.
As an optional implementation manner, step S304, retrieving object feature information of the recommended object includes: determining identification information of a recommended object; object feature information associated with the identification information is retrieved.
In this embodiment, a recommended object for which a recommendation operation is required to be performed is determined, each recommended object has its corresponding identification information, the identification information of the recommended object is determined, and object feature information associated with the identification information is called based on the identification information. The identification information may be an identity ID of the recommended object, and is used for determining the identity of the recommended object.
Alternatively, the object feature information is stored in advance, and the object feature information corresponds to the identification information one by one, and after determining the recommended object, the object feature information associated with the identification information may be called from the database based on the identification information of the recommended object.
For example, when the online e-commerce platform needs to recommend a product size to a recommended object, a request may be sent to the scoring service, and a list of candidate product SKUs that need to be scored and identification information of the recommended object may be placed in the request. After the scoring service accepts the request, the request may be parsed to determine identification information, and the candidate product set, object feature information of the recommended object may be retrieved from the database based on the identification information. Wherein the candidate product SKU list may be a set of candidate products.
As an alternative embodiment, the object characteristic information includes at least one of: size information, historical purchase sequence, user portrayal.
In this embodiment, the object characteristic information may include at least one of: the size information, the historical purchase sequence and the user portrayal, it should be noted that the content of the object feature information is not particularly limited herein, and the information that can be used to characterize the recommended object feature should be within the scope of protection of the present application.
Alternatively, the size information may be information in a user size file, and may be used to determine information such as height, weight, etc. of the recommended object. The historical purchase sequence may be a purchase sequence of the recommended object in the historical event segment, and the sequence includes information of products, sellers, shops, brands, product sizes and the like. The user image can be used for representing information such as the sex, age bracket, labor consumption, common goods receiving address and the like of the user.
In the embodiment, the product recommended to the recommended object is determined based on the characteristics of multiple dimensions of the recommended object, but not just the recommended object is recommended to the recommended object based on the size information of the user, so that the technical effect of improving the accuracy of product recommendation is achieved, and the technical problem of low accuracy of product recommendation is solved.
As an optional implementation manner, step S310, determining at least one product to be recommended from the candidate product set based on the recommendation information, includes: sorting the candidate products according to the recommendation information to obtain a sorting result; and determining the candidate products ranked in the ranking result and positioned in front of the target ranking as target candidate products.
In this embodiment, the recommendation degrees of the plurality of candidate products may be determined according to the recommendation information, and the plurality of candidate products may be ranked based on the recommendation degrees of the candidate products to obtain the ranking result. The candidate products ranked in the ranking result that are located before the target ranking may be determined as target candidate products.
Alternatively, the recommendation information is determined, if the recommendation degree is higher, the recommendation object is indicated to have higher possibility of selecting the product, and if the recommendation degree is lower, the recommendation object is indicated to have lower possibility of selecting the product, and the product is not recommended.
In the embodiment of the application, the object characteristic information of the recommended object is determined, the object characteristic information is analyzed by using the recommendation model to determine the recommendation information corresponding to different candidate products, and the target candidate product is determined based on the recommendation information, so that the technical effect of improving the accuracy of product recommendation is achieved, and the technical problem of low accuracy of product recommendation is solved.
Aiming at the training of the recommended model, the embodiment of the application also provides a training method of the model. Fig. 4 is a flowchart of a method of training a model according to an embodiment of the present application, as shown in fig. 4, which may include the following steps.
Step S402, obtaining historical object feature information of a historical recommended object and historical attribute information of historical products in an electronic commerce platform, wherein the historical object feature information is used for at least representing preference degrees of the historical recommended object on different types of historical products to be transacted.
In the technical scheme provided in the above step S402 of the present application, a recommendation model may be obtained through training, and before training the recommendation model, training data of the recommendation model needs to be obtained, and historical object feature information of a historical recommendation object and historical attribute information of a historical product in an e-commerce platform may be obtained. And training to obtain a recommendation model by taking the characteristic information and the historical attribute information of the historical object as training data. The recommendation model is used for determining products to be recommended to the recommended objects. The historical object characteristic information may be used to characterize at least the degree of preference of the historical recommendation object for different types of historical products to be transacted. The historical recommendation objects may be recommendation objects in a historical event segment. The historical product may be a product purchased by the historical recommendation object.
And step S404, training the target attention model by utilizing the characteristic information and the history attribute information of the history object to obtain a recommendation model.
In the technical solution provided in step S404 of the present application, the target attention model may be trained according to the incremental learning manner by using the historical object feature information and the historical attribute information, so as to obtain the recommendation model.
As an optional implementation manner, step S404 trains the target attention model by using the historical object feature information and the historical attribute information to obtain a recommendation model, which includes: based on the historical purchasing sequence in the historical object characteristic information, constructing and obtaining a training sample; and training the target attention model by using the training sample and the historical attribute information to obtain a recommended model.
In this embodiment, a historical purchase sequence in the historical object feature is obtained, and a training sample including positive and negative samples may be constructed based on the historical purchase sequence, for example, when the historical recommended object purchases a garment one with a size of a middle code, the middle size of the garment one is the positive sample, and other sizes are negative samples, so that the training sample is constructed based on the historical purchase sequence. And training the target attention model by utilizing the positive and negative samples and the historical attribute information to obtain a recommended model.
Optionally, constructing to obtain positive and negative samples, analyzing the historical attribute information and the historical object characteristic information by using the target attention model, determining a recommendation result, determining a loss function corresponding to the recommendation result based on the positive and negative samples, and adjusting model parameters of the target attention model based on the loss function, thereby obtaining the recommendation model.
As an alternative embodiment, the method further comprises: responding to the obtained recommendation instruction, and calling a recommendation model; and processing the object characteristic data by using the recommendation model, and determining a candidate product set and recommendation information of different candidate products in the candidate product set.
In this embodiment, when a recommendation operation needs to be performed on the recommended object, a recommendation instruction may be issued to control the recommendation model to determine a product to be recommended. The recommendation system responds to the obtained recommendation instruction, can call a recommendation model, processes object feature data by using the recommendation model to determine a candidate product set and recommendation information of different candidate products in the candidate product set, determines at least one product to be recommended from the candidate product set based on the recommendation information, and wants a recommendation object to recommend the product to be recommended.
In the embodiment of the application, the historical object characteristic information of the historical recommended object and the historical attribute information of the historical products in the electronic commerce platform are obtained, wherein the historical object characteristic information is used for at least representing the preference degree of the historical recommended object on different types of historical products to be transacted; training a target attention model by utilizing the historical object characteristic information and the historical attribute information to obtain a recommendation model, determining recommendation information of different candidate products by utilizing the trained recommendation model so as to determine at least one product to be recommended, thereby realizing the technical effect of improving the accuracy of product recommendation and solving the technical problem of low accuracy of product recommendation.
According to the embodiment of the application, a recommendation method of products is further provided aiming at the scenes of the augmented reality technology (Augmented Real ity, abbreviated as AR) and the Virtual reality technology (VR). FIG. 5 is a flow chart of another product recommendation method according to an embodiment of the present application. As shown in fig. 5, the method may include the steps of:
step S502, displaying a recommended object which needs to execute a recommending operation on a display screen of an operation interface.
In the technical solution provided in step S502 of the present application, a recommended object that needs to execute a recommendation operation may be displayed on a presentation screen of an operation interface. The operation interface may be a display interface of the mobile terminal, or an interactive interface of the virtual reality VR device or the augmented reality AR device, and the type of the operation interface is not specifically limited herein.
Step S504, determining and displaying the candidate product set and/or at least one product to be recommended in the candidate product set in response to the recommendation command acted on the operation interface.
In the technical solution provided in step S504, when a recommendation operation needs to be performed on a recommended object, a recommendation instruction may be issued by clicking a control on an operation interface. In response to a recommendation instruction acting on the operation interface, feature information of a recommendation object can be analyzed by using a recommendation model, a candidate product set is determined, recommendation information of different candidate products in the candidate product set can be displayed on the operation interface, and/or at least one product to be recommended in the candidate product set can be displayed on the operation interface.
Optionally, when a recommending operation needs to be performed on the recommending object, the recommending object needing to be performed on the presenting picture of the operation interface may be displayed, and a recommending instruction may be issued at the same time, the recommending instruction acting on the operation interface is responded, the object feature information is analyzed by using a recommending model, a candidate product set is determined, and recommending information of different candidate products in the candidate product set is determined, at least one product to be recommended is determined from the candidate product set based on the recommending information, and the recommending object in the presenting picture is provided with the product to be recommended, so as to display an avatar of the recommending object in the virtual world. The recommended object may determine a wearing effect of the product to be recommended according to the avatar, thereby determining whether to purchase the product to be recommended.
Alternatively, the recommendation instructions on the operating interface of the virtual reality device or the augmented reality device may be triggered by the voice acquisition device or the user, and in response to the recommendation instructions, the product to be recommended is determined.
In this embodiment, the product recommendation method may be applied to a hardware environment formed by a server and a virtual reality device. The server may be a server corresponding to the media file operator by sending a recommendation command through the VR device or the AR device, where the network includes but is not limited to: the virtual reality device is not limited to a wide area network, a metropolitan area network, or a local area network: virtual reality helmets, virtual reality glasses, virtual reality all-in-one machines, and the like.
Optionally, the virtual reality device may include: memory, processor, and transmission means. The memory is used to store an application program that can be used to perform: determining a recommended object which needs to execute a recommending operation; object feature information of the recommended object is called, wherein the object feature information is used for at least representing the preference degree of the recommended object for different types of products; analyzing object feature information by using a recommendation model, determining a candidate product set and recommendation information of different candidate products in the candidate product set, wherein the recommendation information is used for predicting the probability that a recommendation object performs transaction operation on the candidate products; determining at least one product to be recommended from the candidate product set based on the recommendation information; and recommending the product to be recommended to the recommended object.
It should be noted that, the method for recommending a product applied to the VR device or the AR device according to this embodiment may include the method of the embodiment shown in fig. 3, so as to achieve the purpose of driving the VR device or the AR device to display the avatar of the recommended object in the virtual world.
Alternatively, the processor of this embodiment may call the application program stored in the memory through the transmission device to perform the above steps. The transmission device can receive the media file sent by the server through the network and can also be used for data transmission between the processor and the memory.
Optionally, in the virtual reality device, a head mounted display with eye tracking is provided, a screen in the head mounted display of the HMD is used for displaying a video picture displayed, an eye tracking module in the HMD is used for acquiring a real-time motion path of an eye of a user, a tracking system is used for tracking position information and motion information of the user in a real three-dimensional space, a calculation processing unit is used for acquiring the real-time position and motion information of the user from the tracking system, calculating three-dimensional coordinates of the head of the user in the virtual three-dimensional space, and visual field orientation of the user in the virtual three-dimensional space.
In this embodiment of the present application, the virtual reality device may be connected to a terminal, where the terminal and the server are connected through a network, and the virtual reality device is not limited to: the terminal is not limited to a PC, a mobile phone, a tablet PC, etc., and the server may be a server corresponding to a media file operator, and the network includes but is not limited to: a wide area network, a metropolitan area network, or a local area network.
Fig. 6 is a schematic view of a recommendation result of a product according to an embodiment of the present application, and shows an avatar of a recommended object in a virtual world by responding to a recommendation command acting on an operation interface, as shown in fig. 6. Wherein the avatar may be an avatar including a product to be recommended worn.
According to the embodiment of the application, a recommended object which needs to execute a recommended operation is displayed on a display screen of an operation interface; responding to a recommendation instruction acted on an operation interface, and determining and displaying a candidate product set and/or at least one product to be recommended in the candidate product set; the candidate product set is generated by analyzing the characteristic information of the recommended object by using the recommendation model, the characteristic information of the recommended object is also used for determining the recommendation information of different candidate products in the candidate product set, and the recommendation information is used for screening out the products to be recommended from the candidate product set, so that the technical effect of improving the accuracy of the recommendation of the candidate products is realized, and the technical problem of low accuracy of the recommendation of the products is solved.
Example 2
Currently, in an electronic marketplace, it is desirable to recommend an appropriate size for a product selected by a consumer. In the related art, a product to be recommended is generally determined based on a product size table and a size file of a consumer. In the method, the accuracy of the product rule code table is highly dependent on the proportion and the accuracy of the rule code table uploaded by merchants, but the proportion and the corresponding accuracy of the rule code table uploaded by middle and small merchants are low, so that the recommendation accuracy is low.
Meanwhile, the product size table describes objective properties of the product, such as information of three circumferences, shoulder widths and the like, and a certain gap exists between the value and the three circumference information of the self-body of the recommended object, so that accurate matching is difficult to perform; and portions of apparel are flexible, where the deviations in the match may be further amplified, resulting in recommended size inaccuracies.
In order to solve the problems, the application provides a personalized clothing size recommendation method based on the stature and historical purchasing behavior of a user, according to the type of a product, size files of recommendation objects with different sexes are automatically used as characteristics, similarity between each size of a candidate product and historical purchasing products of the user is scored by using a recommendation model, dressing preference of the recommendation objects which are bigger or smaller is implicitly considered, the technical effect of improving the accuracy of product recommendation is achieved, and the technical problem of low accuracy of product recommendation is solved.
In the embodiment, under the condition that characteristics of shops, brands, user images and the like are considered, the matching degree between different candidate products and recommended objects in the candidate product set can be scored, the application range of the recommended model is improved, and the technical problem of low accuracy of product recommendation caused by low coverage rate and accuracy of a product rule code table is avoided.
The personalized clothing size recommendation method based on the stature and the historical purchasing behavior of the user is further introduced below.
Fig. 7 is a schematic diagram of recommendation model construction according to an embodiment of the present application, as shown in fig. 7, information such as a user portrait 701, a size file 702, a historical purchasing sequence 703, a commodity SKU 704 and the like can be obtained by using feature engineering, and the deep learning model 707 based on the target attention frame is subjected to incremental learning by using the information and positive and negative samples, so as to obtain a recommendation model 708. The recommendation model may be a size recommendation model or a product type recommendation model, and the type of the recommendation model is not particularly limited herein. The historical purchase sequence may be a purchase sequence of the recommended object in the historical event segment, and the sequence includes information of a product, a seller, a store, a brand, a product size, and the like, for example, the purchase sequence of the recommended object in the past two years may be used. The user image can be used for representing information such as the sex, age bracket, labor consumption, common goods receiving address and the like of the user. The size file may be a size file under a recommended user of a different gender selected according to the gender attribute of the product, for example, a size file of a male of a selected user of a men's suit.
Alternatively, the commodity SKU may include at least one of the following features: identification information, size attributes, brands, candidate product identity information (Ident ity Document, abbreviated as ID), store information, SKU ID, etc., are given here by way of example only, and the contents of the commodity SKU are not particularly limited.
Optionally, the order data of the full platform user is obtained to obtain the full volume order 705, and a training sample 706 including positive and negative samples may be constructed based on the full volume order 705, for example, the first history recommended object purchases the first garment with the size of the middle size, the first garment with the middle size is the positive sample, and the other sizes are the negative samples, so as to generate the positive and negative samples.
Optionally, constructing to obtain positive and negative samples, analyzing the historical attribute information and the historical object characteristic information by using the target attention model, determining a recommendation result, determining a loss function corresponding to the recommendation result based on the positive and negative samples, and adjusting model parameters of the target attention model based on the loss function, thereby obtaining the recommendation model.
In the embodiment, the product recommended to the recommended object is determined based on the characteristics of multiple dimensions of the recommended object, but not just the recommended object is recommended to the recommended object based on the size information of the user, so that the technical effect of improving the accuracy of product recommendation is achieved, and the technical problem of low accuracy of product recommendation is solved.
In this embodiment, when a product size needs to be recommended to a recommended object online, a request may be sent to the scoring service of the algorithm, placing a list of product SKUs that need to be scored into the request.
For example, the algorithm deploys the recommendation model as an online service, and the server may trigger the request. For example, when a recommended object enters a front-end detail page of a piece of clothing, the back-end, i.e. the server, requests the algorithm to obtain a recommended result of the clothing size, and displays the recommended result at the front-end.
Fig. 8 is a schematic diagram of determining a recommendation result according to an embodiment of the present application, and as shown in fig. 8, the scoring service determines information such as a user portrait 801, a size file 802, a historical purchasing sequence 803, a commodity SKU 804, and the like according to identification information of a recommendation object, and when the recommendation object needs a size recommendation service, a request 805 may be initiated. In response to the request 805, the recommendation model 806 performs an online scoring service to obtain recommendation scores for each product sku in the candidate product set under the recommendation object. The candidate product 807 with the highest score may be selected, or it may be determined whether the recommendation score is greater than a threshold, e.g., whether the recommendation score is greater than 0.5, and if so, the candidate product of that size is recommended, otherwise, the recommended subject is considered to be more fuzzy to select under that product and not recommended.
In the embodiment, modeling is performed by using a deep learning technology to introduce the characteristics of each dimension of a user and a product, a recommendation model is used, a user history purchasing sequence is fully utilized, understanding of the user is enhanced, and the dressing size preference of the user is considered; meanwhile, through feature screening, the problem of uncontrollable product size table coverage rate and accuracy is avoided, universality of a scheme is greatly improved, the technical effect of improving accuracy of product recommendation is achieved, and the technical problem of product recommendation accuracy is solved.
The method embodiment provided in embodiment 1 of the present application may be executed in a mobile terminal, a computer terminal or a similar computing device. Fig. 9 is a block diagram of a hardware structure of a computer terminal (or mobile device) of a recommended method of an article according to an embodiment of the present application, as shown in fig. 9, the computer terminal 90 (or mobile device) may include one or more (shown by 902a, 902b, … …,902n in the drawing) processors 902 (the processors 902 may include, but are not limited to, a microprocessor (Microcontrol ler Unit, abbreviated as MCU) or a programmable logic device (Field Programmable Gate Array, abbreviated as FPGA) or the like processing means), a memory 904 for storing data, and a transmission means 906 for a communication function. In addition, the method may further include: a display, an input/output interface (I/O interface), a universal serial BUS (Universal Serial Bus, simply USB) port (which may be included as one of the ports of the BUS), a network interface, a power supply, and/or a camera. It will be appreciated by those skilled in the art that the configuration shown in fig. 9 is merely illustrative and is not intended to limit the configuration of the electronic device. For example, the computer terminal 90 may also include more or fewer components than shown in FIG. 9, or have a different configuration than shown in FIG. 9.
The hardware block diagram shown in fig. 9 may be used not only as an exemplary block diagram of the computer terminal 90 (or mobile device) described above, but also as an exemplary block diagram of the server described above, and in an alternative embodiment, fig. 2 shows in block diagram form an embodiment in which the computer terminal 90 (or mobile device) shown in fig. 9 described above is used as a computing node in a computing environment 201.
The memory 904 may be used to store software programs and modules of application software, such as program instructions/data storage devices corresponding to the data processing methods in the embodiments of the present application, and the processor executes the software programs and modules stored in the memory 904, thereby executing various functional applications and data processing, that is, implementing the data processing methods described above. The memory 904 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 904 may further include memory located remotely from the processor, which may be connected to the computer terminal 90 via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The transmission means 906 is used for receiving or transmitting data via a network. The specific examples of the network described above may include a wireless network provided by a communication provider of the computer terminal 90. In one example, the transmission means 906 comprises a network adapter (Network Interface Control ler, NIC) that can be connected to other network devices via a base station to communicate with the internet. In one example, the transmission device 906 may be a Radio Frequency (RF) module for communicating with the internet wirelessly.
The display may be, for example, a touch screen type Liquid Crystal Display (LCD) that may enable a user to interact with a user interface of the computer terminal 90 (or mobile device).
It should be noted that, the object information (including, but not limited to, object device information, object personal information, etc.) and the data (including, but not limited to, data for analysis, stored data, presented data, etc.) related to the present application are information and data authorized by the object or sufficiently authorized by each party, and the collection, use and processing of the related data need to comply with the related laws and regulations and standards of the related country and region, and provide a corresponding operation entry for the object to select authorization or rejection.
It should be noted that, for simplicity of description, the foregoing method embodiments are all expressed as a series of action combinations, but it should be understood by those skilled in the art that the present application is not limited by the order of actions described, as some steps may be performed in other order or simultaneously in accordance with the present application. Further, those skilled in the art will also appreciate that the embodiments described in the specification are all preferred embodiments, and that the acts and modules referred to are not necessarily required in the present application.
From the description of the above embodiments, it will be clear to a person skilled in the art that the method according to the above embodiments may be implemented by means of software plus a necessary general hardware platform, but that it may also be implemented by means of hardware. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk), comprising several instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to perform the method of the embodiments of the present application.
Example 3
According to the embodiment of the application, a product recommending device for implementing the product recommending method shown in fig. 3 is also provided.
Fig. 10 is a schematic diagram of a product recommendation device according to an embodiment of the present application, and as shown in fig. 10, a product recommendation device 1000 may include: a first determination unit 1002, a call unit 1004, a second determination unit 1006, a third determination unit 1008, and a recommendation unit 1010.
A first determining unit 1002, configured to determine a recommendation object that needs to perform a recommendation operation.
The retrieving unit 1004 is configured to retrieve object feature information of a recommended object, where the object feature information is used to at least characterize a preference degree of the recommended object for different types of products.
A second determining unit 1006, configured to analyze the object feature information using the recommendation model, determine a candidate product set, and recommendation information of different candidate products in the candidate product set, where the recommendation information is used to predict a probability that the recommendation object performs a transaction operation on the candidate products.
And a third determining unit 1008, configured to determine at least one product to be recommended from the candidate product set based on the recommendation information.
A recommending unit 910, configured to recommend a product to be recommended to a recommendation object.
Here, the above-described first determination unit 1002, the calling unit 1004, the second determination unit 1006, the third determination unit 1008, and the recommendation unit 1010 correspond to steps S302 to S310 in embodiment 1, and the five units are the same as the examples and application scenarios implemented by the corresponding steps, but are not limited to those disclosed in embodiment 1 above. It should be noted that the above-mentioned units may be hardware components or software components stored in a memory (e.g., the memory 904) and processed by one or more processors (e.g., the processors 902a,902b … …,902 n), or may be part of an apparatus and may be run in the computer terminal 90 provided in embodiment 2.
According to the embodiment of the application, a product recommending device for implementing the product recommending method shown in fig. 4 is also provided.
Fig. 11 is a schematic diagram of a training apparatus for a model according to an embodiment of the present application, as shown in fig. 11, a training apparatus 1100 for the model may include: an acquisition unit 1102 and a first processing unit 1104.
The obtaining unit 1102 is configured to obtain historical object feature information of a historical recommended object and historical attribute information of a historical product in an e-commerce platform, where the historical object feature information is used to at least characterize preference degrees of the historical recommended object to different types of historical products to be transacted.
The first processing unit 1104 is configured to train the target attention model by using the historical object feature information and the historical attribute information to obtain a recommendation model.
Here, it should be noted that the above-mentioned acquisition unit 1102 and the first processing unit 1104 correspond to step S402 to step S404 in embodiment 1, and the two units are the same as the examples and application scenarios implemented by the corresponding steps, but are not limited to those disclosed in embodiment 1 above. It should be noted that the above-mentioned units may be hardware components or software components stored in a memory (e.g., the memory 904) and processed by one or more processors (e.g., the processors 902a,902b … …,902 n), or may be part of an apparatus and may be run in the computer terminal 90 provided in embodiment 2.
FIG. 12 is a schematic diagram of another product recommendation device according to an embodiment of the present application, as shown in FIG. 12, the product recommendation 1200 may include: a display unit 1202 and a second processing unit 1204.
And a display unit 1202, configured to display a recommendation object that needs to perform a recommendation operation on a display screen of the operation interface.
The second processing unit 1204 is configured to determine and display the candidate product set and/or at least one product to be recommended in the candidate product set in response to a recommendation instruction acting on the operation interface; the candidate product set is generated by analyzing feature information of a recommendation object by using a recommendation model, the feature information of the recommendation object is also used for determining recommendation information of different candidate products in the candidate product set, and the recommendation information is used for screening products to be recommended from the candidate product set.
Here, it should be noted that the above-mentioned display unit 1202 and the second processing unit 1204 correspond to step S502 to step S504 in embodiment 1, and the two units are the same as the examples and application scenarios implemented by the corresponding steps, but are not limited to those disclosed in embodiment 1 above. It should be noted that the above-mentioned units may be hardware components or software components stored in a memory (e.g., the memory 904) and processed by one or more processors (e.g., the processors 902a,902b … …,902 n), or may be part of an apparatus and may be run in the computer terminal 90 provided in embodiment 2.
In the recommending device of the product, the object characteristic information of the recommending object is determined, the recommending model is used for analyzing the object characteristic information to determine recommending information corresponding to different candidate products, and the target candidate product is determined based on the recommending information, so that the technical effect of improving the accuracy of product recommendation is achieved, and the technical problem of low accuracy of product recommendation is solved.
Example 4
Embodiments of the present application may provide a computer terminal, which may be any one of a group of computer terminals. Alternatively, in the present embodiment, the above-described computer terminal may be replaced with a terminal device such as a mobile terminal.
Alternatively, in this embodiment, the above-mentioned computer terminal may be located in at least one network device among a plurality of network devices of the computer network.
In this embodiment, the computer terminal may execute the program code of the following steps in the product recommendation method: determining a recommended object which needs to execute a recommending operation; retrieving object feature information of a recommended object, wherein the object feature information is used for at least representing preference degrees of the recommended object on different types of products; analyzing object feature information by using a recommendation model, determining a candidate product set and recommendation information of different candidate products in the candidate product set, wherein the recommendation information is used for predicting the probability that a recommendation object performs transaction operation on the candidate products; determining at least one product to be recommended from the candidate product set based on the recommendation information; and recommending the product to be recommended to the recommended object.
Alternatively, fig. 13 is a block diagram of a computer terminal according to an embodiment of the present application, and as shown in fig. 13, the computer terminal a may include: one or more (only one is shown) processors 1302, memory 1304, and transmission means 1306.
The memory may be used to store software programs and modules, such as program instructions/modules corresponding to the method and apparatus for recommending a product in the embodiments of the present application, and the processor executes the software programs and modules stored in the memory, thereby executing various functional applications and data processing, that is, implementing the method for recommending a product described above. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory may further comprise memory remotely located from the processor, the remote memory being connectable to the computer terminal a through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The processor may call the information and the application program stored in the memory through the transmission device to perform the following steps: determining a recommended object which needs to execute a recommending operation; retrieving object feature information of a recommended object, wherein the object feature information is used for at least representing preference degrees of the recommended object on different types of products; analyzing object feature information by using a recommendation model, determining a candidate product set and recommendation information of different candidate products in the candidate product set, wherein the recommendation information is used for predicting the probability that a recommendation object performs transaction operation on the candidate products; determining at least one product to be recommended from the candidate product set based on the recommendation information; and recommending the product to be recommended to the recommended object.
Optionally, the above processor may further execute program code for: acquiring attribute information of different candidate products; analyzing object feature information and attribute information of different candidate products by using a recommendation model, and determining recommendation scores of the different candidate products; recommendation information for the different candidate products is determined based on the recommendation scores.
Optionally, the above processor may further execute program code for: analyzing object feature information by using a recommendation model, and determining historical transaction preference of a recommendation object on at least one historical product; analyzing attribute information by using a recommendation model, and determining transaction preferences of a recommendation object on different candidate products; determining a similarity between the historical transaction preferences and the transaction preferences using a recommendation model; and determining a recommendation score of the candidate product based on the similarity, wherein the similarity and the recommendation score are positively correlated.
Optionally, the above processor may further execute program code for: and analyzing a historical transaction sequence in the object characteristic information by using the recommendation model, and determining the historical transaction preference of the recommendation object on at least one historical product, wherein the historical transaction sequence is used for representing the purchase condition of the recommendation object on the at least one historical product.
Optionally, the above processor may further execute program code for: acquiring a historical browsing sequence with the correlation degree with the candidate product higher than a correlation degree threshold value in the object characteristic information, wherein the historical browsing sequence is used for representing the browsing condition of a recommended object on at least one historical product in a target time period; the historical browsing sequence is analyzed using the recommendation model to determine historical transaction preferences for at least one historical product.
Optionally, the above processor may further execute program code for: using a recommendation model to determine a historical product with the same type as the candidate product in at least one historical product as a target historical product; determining a target historical transaction preference of a target historical product; using the recommendation model, a similarity between the target historical transaction preferences and the transaction preferences of the different candidate products is determined.
Optionally, the above processor may further execute program code for: determining the size in the attribute information, the historical purchase sequence in the object feature information and the size information in the object feature information; determining a first similarity between the size and the size information of the candidate product using the recommendation model, and determining a second similarity between the candidate product and the historical product in the historical purchase sequence using the recommendation model; weighting the first similarity and the second similarity to obtain the similarity; based on the similarity, a recommendation score for the candidate product is determined.
Optionally, the above processor may further execute program code for: determining identification information of a recommended object; object feature information associated with the identification information is retrieved.
The processor may call the information and the application program stored in the memory through the transmission device to perform the following steps: acquiring historical object characteristic information of a historical recommended object and historical attribute information of a historical product in an electronic commerce platform, wherein the historical object characteristic information is used for at least representing preference degrees of the historical recommended object on different types of historical products to be transacted; and training the target attention model by utilizing the historical object characteristic information and the historical attribute information to obtain a recommendation model.
The processor may call the information and the application program stored in the memory through the transmission device to perform the following steps: displaying a recommended object which needs to execute a recommended operation on a display picture of an operation interface; responding to a recommendation instruction acted on an operation interface, and determining and displaying a candidate product set and/or at least one product to be recommended in the candidate product set; the candidate product set is generated by analyzing feature information of a recommendation object by using a recommendation model, the feature information of the recommendation object is also used for determining recommendation information of different candidate products in the candidate product set, and the recommendation information is used for screening products to be recommended from the candidate product set.
By adopting the embodiment of the application, the object characteristic information of the recommended object is determined, the recommendation model is used for analyzing the object characteristic information so as to determine the recommendation information corresponding to different candidate products, and the target candidate product is determined based on the recommendation information, so that the technical effect of improving the accuracy of product recommendation is achieved, and the technical problem of low accuracy of product recommendation is solved.
It will be understood by those skilled in the art that the structure shown in fig. 13 is only schematic, and the computer terminal a may also be a terminal device such as a smart phone (e.g. an Android phone, an iOS phone, etc.), a tablet computer, a palm computer, and a mobile internet device (Mobi le Internet Devices, abbreviated as MID), a PAD, etc. Fig. 13 does not limit the structure of the computer terminal a. For example, the computer terminal a may also include more or fewer components (such as a network interface, a display device, etc.) than shown in fig. 13, or have a different configuration than shown in fig. 13.
Those of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be implemented by a program for instructing a terminal device to execute in association with hardware, the program may be stored in a computer readable storage medium, and the storage medium may include: flash disk, read-Only Memory (ROM), random access Memory (Random Access Memory, RAM), magnetic disk or optical disk, etc.
Example 5
Embodiments of the present application also provide a computer-readable storage medium. Alternatively, in this embodiment, the computer readable storage medium may be used to store the program code executed by the recommendation method for the product provided in the first embodiment.
Alternatively, in this embodiment, the above-mentioned computer-readable storage medium may be located in any one of the computer terminals in the computer terminal group in the computer network, or in any one of the mobile terminals in the mobile terminal group.
Optionally, in the present embodiment, the computer readable storage medium is configured to store program code for performing the steps of: determining a recommended object which needs to execute a recommending operation; retrieving object feature information of a recommended object, wherein the object feature information is used for at least representing preference degrees of the recommended object on different types of products; analyzing object feature information by using a recommendation model, determining a candidate product set and recommendation information of different candidate products in the candidate product set, wherein the recommendation information is used for predicting the probability that a recommendation object performs transaction operation on the candidate products; determining at least one product to be recommended from the candidate product set based on the recommendation information; and recommending the product to be recommended to the recommended object.
Optionally, the above computer readable storage medium may further execute program code for: acquiring attribute information of different candidate products; analyzing object feature information and attribute information of different candidate products by using a recommendation model, and determining recommendation scores of the different candidate products; recommendation information for the different candidate products is determined based on the recommendation scores.
Optionally, the above computer readable storage medium may further execute program code for: analyzing object feature information by using a recommendation model, and determining historical transaction preference of a recommendation object on at least one historical product; analyzing attribute information by using a recommendation model, and determining transaction preferences of a recommendation object on different candidate products; determining a similarity between the historical transaction preferences and the transaction preferences using a recommendation model; and determining a recommendation score of the candidate product based on the similarity, wherein the similarity and the recommendation score are positively correlated.
Optionally, the above computer readable storage medium may further execute program code for: and analyzing a historical transaction sequence in the object characteristic information by using the recommendation model, and determining the historical transaction preference of the recommendation object on at least one historical product, wherein the historical transaction sequence is used for representing the purchase condition of the recommendation object on the at least one historical product.
Optionally, the above computer readable storage medium may further execute program code for: acquiring a historical browsing sequence with the correlation degree with the candidate product higher than a correlation degree threshold value in the object characteristic information, wherein the historical browsing sequence is used for representing the browsing condition of a recommended object on at least one historical product in a target time period; the historical browsing sequence is analyzed using the recommendation model to determine historical transaction preferences for at least one historical product.
Optionally, the above computer readable storage medium may further execute program code for: using a recommendation model to determine a historical product with the same type as the candidate product in at least one historical product as a target historical product; determining a target historical transaction preference of a target historical product; using the recommendation model, a similarity between the target historical transaction preferences and the transaction preferences of the different candidate products is determined.
Optionally, the above computer readable storage medium may further execute program code for: determining the size in the attribute information, the historical purchase sequence in the object feature information and the size information in the object feature information; determining a first similarity between the size and the size information of the candidate product using the recommendation model, and determining a second similarity between the candidate product and the historical product in the historical purchase sequence using the recommendation model; weighting the first similarity and the second similarity to obtain the similarity; based on the similarity, a recommendation score for the candidate product is determined.
Optionally, the above computer readable storage medium may further execute program code for: determining identification information of a recommended object; object feature information associated with the identification information is retrieved.
As an alternative example, the computer readable storage medium is arranged to store program code for performing the steps of: acquiring historical object characteristic information of a historical recommended object and historical attribute information of a historical product in an electronic commerce platform, wherein the historical object characteristic information is used for at least representing preference degrees of the historical recommended object on different types of historical products to be transacted; and training the target attention model by utilizing the historical object characteristic information and the historical attribute information to obtain a recommendation model.
As an alternative example, the computer readable storage medium is arranged to store program code for performing the steps of: displaying a recommended object which needs to execute a recommended operation on a display picture of an operation interface; responding to a recommendation instruction acted on an operation interface, and determining and displaying a candidate product set and/or at least one product to be recommended in the candidate product set; the candidate product set is generated by analyzing feature information of a recommendation object by using a recommendation model, the feature information of the recommendation object is also used for determining recommendation information of different candidate products in the candidate product set, and the recommendation information is used for screening products to be recommended from the candidate product set.
In the embodiment of the application, the object characteristic information of the recommended object is determined, the object characteristic information is analyzed by using the recommendation model to determine the recommendation information corresponding to different candidate products, and the target candidate product is determined based on the recommendation information, so that the technical effect of improving the accuracy of product recommendation is achieved, and the technical problem of low accuracy of product recommendation is solved.
Example 6
Embodiments of the present application may provide an electronic device that may include a memory and a processor.
Fig. 14 is a block diagram of an electronic device of a product recommendation method according to an embodiment of the present application. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the application described and/or claimed herein.
As shown in fig. 14, the apparatus 1400 includes a computing unit 1401 that can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM) 1402 or a computer program loaded from a storage unit 1408 into a random access Memory (Random Access Memory, RAM) 1403. In the RAM1403, various programs and data required for the operation of the device 1400 can also be stored. The computing unit 1401, the ROM1402, and the RAM1403 are connected to each other through a bus 1404. An input/output (I/O) interface 1405 is also connected to the bus 1404.
Various components in device 1400 are connected to I/O interface 1405, including: an input unit 1406 such as a keyboard, a mouse, or the like; an output unit 1404 such as various types of displays, speakers, and the like; a storage unit 1408 such as a magnetic disk, an optical disk, or the like; and a communication unit 1409 such as a network card, a modem, a wireless communication transceiver, and the like. The communication unit 1409 allows the device 1400 to exchange information/data with other devices through a computer network such as the internet and/or various telecommunications networks.
The computing unit 1401 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 1401 include, but are not limited to, a central processing unit (Central Processing Unit, abbreviated as CPU), a graphics processing unit (Graphics Processing Unit, abbreviated as GPU), various dedicated artificial intelligence (Artificial Intelligence, abbreviated as AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (Demand Side Platform, abbreviated as DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 1401 performs the respective methods and processes described above, for example, a verification method of data. For example, in some embodiments, the method of verifying data may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 1408. In some embodiments, part or all of the computer program may be loaded and/or installed onto the device 1400 via the ROM1402 and/or the communication unit 1409. When a computer program is loaded into the RAM1403 and executed by the computing unit 1401, one or more steps of the verification method of data described above may be performed. Alternatively, in other embodiments, the computing unit 1401 may be configured to perform the verification method of the data by any other suitable means (e.g. by means of firmware).
According to an embodiment of the present application, there is provided a recommendation method for an article, it being noted that the steps shown in the flowchart of the drawings may be performed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be performed in an order different from that herein.
Various implementations of the systems and techniques described here above can be implemented in digital electronic circuitry, integrated circuit systems, field programmable gate arrays (Field Programmable Gate Array, abbreviated as FPGAs), application specific integrated circuits (Application Specific Integrated Circuit, abbreviated as ASICs), application specific standard products (Application Specific Standard Product, abbreviated as ASSPs), systems on a chip (SOC), complex programmable logic devices (Complex Programmable Logic Device, abbreviated as CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.
Program code for carrying out methods of the present application may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of this application, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: other types of devices may also be used to provide interaction with a user, for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback), and input from the user may be received in any form (including acoustic input, speech input, or tactile input).
The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local area network (Local Area Network, abbreviated LAN), wide area network (Wide Area Network, abbreviated WAN) and the internet.
The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server may be a cloud server, a server of a distributed system, or a server incorporating a blockchain.
It should be noted that, the foregoing embodiment numbers are merely for describing the embodiments, and do not represent the advantages and disadvantages of the embodiments.
In the foregoing embodiments of the present application, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.
In the several embodiments provided in the present application, it should be understood that the disclosed technology content may be implemented in other manners. The above-described embodiments of the apparatus are merely exemplary, and are merely a logical functional division, and there may be other manners of dividing the apparatus in actual implementation, for example, multiple units or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interfaces, units or modules, or may be in electrical or other forms.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in each embodiment of the present application may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be embodied in essence or a part contributing to the prior art or all or part of the technical solution, in the form of a software product stored in a storage medium, including several instructions to cause a computer device (which may be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the methods of the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a removable hard disk, a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The foregoing is merely a preferred embodiment of the present application and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present application and are intended to be comprehended within the scope of the present application.