Detailed Description
The present application will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present application more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
It should be noted that although functional block division is performed in a device diagram and a logic sequence is shown in a flowchart, in some cases, the steps shown or described may be performed in a different order than the block division in the device, or in the flowchart. The terms first, second and the like in the description and in the claims and in the above-described figures, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of the application only and is not intended to be limiting of the application.
First, several nouns involved in the present application are parsed:
The graph structure data graph structured data is a data structure representing relationships between entities and attributes of the entities themselves. Graph structure data is an effective method of handling complex relational networks in computer science. In the graph structure data, the data is represented as nodes and edges. The graph data may simulate various types of networks, such as social networks, knowledge graphs, traffic networks, and the like. The processing of graph structure data involves theory and methodology of graph theory, an important branch in the fields of data science, machine learning and artificial intelligence. The main applications include network analysis, recommendation system, path optimization, pattern recognition, etc. The graph structure data can be used not only to analyze network structure and dynamics, but also to optimize network performance and discover potential information in the data.
The sparse multi-class cross entropy penalty (sparse categorical crossentropy loss) is a penalty function for machine learning classification tasks, and particularly when dealing with multi-class classification problems, can effectively measure the difference between model predictive probability distribution and actual labels. Sparse multi-class cross entropy loss applies to the case where the tag exists in integer form, rather than in one-hot encoding (one-hot encoding). In the loss function calculation process, the real class labels are firstly converted into class probability distribution, and then cross entropy between the prediction probability distribution and the real probability distribution is calculated. The sparse multi-category cross entropy loss is mainly applied to deep learning, and particularly applied to classification models in the fields of image recognition, natural language processing, voice recognition and the like. It not only helps to increase computational efficiency, but also effectively addresses the problem of class imbalance in large-scale data sets. By minimizing this loss, the model can better learn the mapping from the input data to the output labels, thereby improving the accuracy of classification and generalization of the model.
Focal loss (focal loss), a loss function widely used in deep learning and machine learning, particularly when dealing with problems with significant class imbalance, such as in some image processing and object detection tasks. Focus loss is a variation of the cross entropy loss function that solves the problem of class imbalance by adjusting the weights of misclassified samples, making the model more focused on difficult-to-classify samples. In the focus loss, by introducing an adjustment factor, the contribution of those samples that are easily classified to the total loss is reduced, thereby making the model more focused on those samples that are difficult to identify. The method not only improves the recognition capability of the model to minority samples, but also enhances the robustness of the model. The focus loss is mainly applied to the field of computer vision, especially in tasks such as object detection and semantic segmentation, and the like, so that the recognition performance and the generalization capability of the model are effectively improved.
Genetic search (GENETIC SEARCH) is a search algorithm that mimics natural selection and genetic principles, and belongs to a branch of evolutionary algorithm. Such algorithms use mechanisms similar to biological genetic and natural selection to solve optimization and search problems. Genetic searching typically involves the processes of selection, crossover, and mutation, through manipulation of these biological heuristics to explore the solution space for optimal solutions to the problem. Genetic search is widely applied to the fields of computer science, engineering optimization, artificial intelligence and the like, and is particularly suitable for solving the problems of excessive complexity or excessive solution space for the traditional search method. The candidate solution is continuously optimized by simulating the evolution process of the population, so that the searching efficiency and the quality of the solution are improved. The main advantage of genetic search is that it is robust and can find globally optimal solutions in complex multimodal problems. In addition, genetic searching can be combined with other searching and optimizing techniques to form a hybrid algorithm to solve more complex problems.
Product recommendation refers to recommending a product to a user, for example, when recommending an insurance product, a target product meeting a target user can be screened out from the insurance product, and then the target product is pushed to the target user. In the scene of insurance product recommendation, insurance recommendation is usually realized based on a neural network, but when the neural network is trained, the recommendation capability of the neural network does not have generalization because the data set cannot cover various conditions of users, so that the product recommendation success rate is not high. Therefore, how to improve the success rate of recommendation of insurance products becomes a problem to be solved.
Based on the above, the embodiment of the application provides a product recommendation method and device, electronic equipment and storage medium, aiming at improving the recommendation success rate of insurance products.
The embodiment of the application provides a product recommendation method and device, electronic equipment and a storage medium, and specifically, the following embodiment is used for explaining, and first describing the product recommendation method in the embodiment of the application.
The embodiment of the application can acquire and process the related data based on the artificial intelligence technology. Wherein artificial intelligence (ARTIFICIAL INTELLIGENCE, AI) is the theory, method, technique, and application system that uses a digital computer or a digital computer-controlled machine to simulate, extend, and expand human intelligence, sense the environment, acquire knowledge, and use knowledge to obtain optimal results.
Artificial intelligence infrastructure technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and other directions.
The product recommendation method provided by the embodiment of the application relates to the technical field of intelligent recommendation, and is suitable for the technical field of finance and technology. The product recommendation method provided by the embodiment of the application can be applied to the terminal, the server side and software running in the terminal or the server side. In some embodiments, the terminal may be a smart phone, a tablet computer, a notebook computer, a desktop computer, etc., the server may be configured as an independent physical server, may be configured as a server cluster or a distributed system formed by a plurality of physical servers, and may be configured as a cloud server for providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDNs, and basic cloud computing services such as big data and artificial intelligent platforms, and the software may be an application for implementing a product recommendation method, but is not limited to the above form.
The application is operational with numerous general purpose or special purpose computer system environments or configurations. Such as a personal computer, a server computer, a hand-held or portable device, a tablet device, a multiprocessor system, a microprocessor-based system, a set top box, a programmable consumer electronics, a network PC, a minicomputer, a mainframe computer, a distributed computing environment that includes any of the above systems or devices, and the like. The application may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The application may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.
It should be noted that, in each specific embodiment of the present application, when related processing is required according to user information, user behavior data, user history data, user location information, and other data related to user identity or characteristics, permission or consent of the user is obtained first, and the collection, use, processing, and the like of the data comply with related laws and regulations and standards. In addition, when the embodiment of the application needs to acquire the sensitive personal information of the user, the independent permission or independent consent of the user is acquired through popup or jump to a confirmation page and the like, and after the independent permission or independent consent of the user is definitely acquired, the necessary relevant data of the user for enabling the embodiment of the application to normally operate is acquired.
Fig. 1 is an optional flowchart of a product recommendation method according to an embodiment of the present application, where the method in fig. 1 may include, but is not limited to, steps S101 to S106.
Step S101, sample user information of a sample object and a sample product purchase record of the sample object are obtained;
step S102, constructing a user portrait according to sample user information to obtain a sample user portrait;
step S103, training a preset product prediction model according to the sample user portrait and the sample product purchase record to obtain a target product prediction model;
Step S104, genetic searching is carried out on the sample product purchase record and the sample user portrait to obtain a selected user portrait;
Step S105, obtaining candidate user information of the candidate object, and screening similar users of the candidate object according to the selected user portrait and the candidate user information to obtain a target object;
and S106, carrying out product prediction on the target object based on the target product prediction model to obtain a target product, and carrying out product recommendation on the target object by the target product.
The method comprises the steps S101 to S106 of the embodiment of the application, wherein the sample user information of a sample object and the sample product purchase record of the sample object are obtained, then user portrayal construction is carried out according to the sample user information to obtain the sample user portrayal, then a preset product prediction model is trained according to the sample user portrayal and the sample product purchase record to obtain a target product prediction model, a model which can accurately predict a user according to the user portrayal is obtained, thereby improving the recommendation success rate of an insurance product, further genetic search is carried out on the sample product purchase record and the sample user portrayal to obtain a selected user portrayal, thereby obtaining more data sets as much as possible, enabling the data sets to cover various user conditions as much as possible, further candidate user information of candidate objects is obtained, and similar user screening is carried out on the candidate objects according to the selected user portrayal and the candidate user information to obtain a target object, thus the target object is found out in a huge customer population, thus accurately determining a user needing to be recommended for the product, thereby improving the recommendation success rate of the insurance product is obtained, finally, the target product is predicted on the target object is based on the target product prediction model, further, more accurate data sets are obtained, and the target product is recommended on the target product is accurately recommended on the target product and the target product is accurately recommended on the user to the user, and the user is accurately recommended, and the recommendation is screened according to the recommendation product is achieved.
In step S101 of some embodiments, the sample object refers to an individual participating in the purchase of the insurance product, including a subject who purchased the insurance product, such as a natural person, business organization, or social group. Sample user information refers to data related to sample objects, such as personal basic information name, gender, age, date of birth and the like, professional information occupation type, industry, work unit name and the like, economic condition related information such as income level, asset condition and the like, past medical history of health condition information, current physical condition, whether familial genetic diseases exist or not and the like. The sample product purchase record is detailed record data of insurance product purchase behavior made by the sample object, including but not limited to, insurance product details including purchase, purchase time, purchase channel, premium payment condition, and the like.
Referring to fig. 2, in some embodiments, step S102 may include, but is not limited to, steps S201 to S203:
step S201, constructing graph structure data according to sample user information to obtain sample user graph structure data;
Step S202, screening a preset edge weight calculation template according to sample user information to obtain a selected edge weight calculation template;
And step S203, carrying out edge weight assignment on the sample user graph structure data based on the selected edge weight calculation template to obtain a sample user graph.
In the steps S201 to S203 shown in the embodiment of the present application, the graph structure data is constructed according to the sample user information to obtain the sample user graph structure data, then the preset edge weight calculation template is screened according to the sample user information to obtain the selected edge weight calculation template, and finally the edge weight assignment is performed on the sample user graph structure data based on the selected edge weight calculation template to obtain the sample user portrait, so that the user characteristics are accurately presented, and the recommendation success rate of the subsequent insurance products is improved.
In step S201 of some embodiments, the sample user graph structure data is a data structure constructed based on sample user information, the sample user is abstracted into a master node, the user information of the sample user is abstracted into sub-nodes of the connection node, and each sub-node is connected with the master node to obtain the sample user graph structure data.
In step S202 of some embodiments, screening refers to screening the selected edge weight calculation template from the preset edge weight calculation template, specifically, according to information included in the sample user information, for example, the preset edge weight calculation template includes ((income×age), (asset-liability), log (income), (income > 50000) AND (liability < 20000), NOT (number of claims to be resolved > 2)), if the sample user information includes only income, age, asset AND liability, AND does NOT include number of claims to be resolved, the selected edge weight calculation template may be screened ((income×age), (asset-liability), log (income), (50000) AND (liability < 20000)), AND if the sample user information includes only age, asset, liability AND number of claims to be resolved, the selected edge weight calculation template may be screened ((asset-liability), AND (liability < 20000)).
In step S203 of some embodiments, edge weight assignment is performed on the sample user graph structure data based on the selected edge weight calculation template, so as to obtain a sample user image, for example, if the sample user information includes only age, asset, liability, AND number of claims AND the edge weight calculation template is selected ((asset-liability), AND (liability < 20000)), the child node formed by the asset AND the child node formed by the liability are connected, AND the value of the corresponding edge is set to (asset-liability). The liability child nodes are self-connected once, AND the corresponding edges are set to AND (liability < 20000).
If the sample user information includes income, age, assets AND liabilities, the number of claims is not included, AND the edge weight calculation template is selected as ((income×age), (asset-liability), log (income), (income > 50000) AND (liability < 20000)), the child nodes formed by the ages AND the child nodes formed by the incomes are connected, AND the corresponding edge is assigned as the corresponding value of (income×age). (asset-liability), log (income), (income > 50000) AND (liability < 20000) are the same AND are not described in detail here.
Referring to fig. 3, in some embodiments, step S103 may include, but is not limited to, steps S301 to S304:
step S301, carrying out graph connection on a sample user graph according to a sample product purchase record to obtain original graph structure data;
Step S302, carrying out weight assignment on the image structure data according to the sample product purchase record, and selecting an image data structure;
step S303, product prediction is carried out on the selected image data structure according to a preset product prediction model and a sample product purchase record, and a predicted product purchase record is obtained;
And step S304, carrying out parameter optimization on the preset product prediction model according to the predicted product purchase record and the sample product purchase record to obtain a target product prediction model.
In the steps S301 to S304 shown in the embodiment of the present application, the sample user portraits are connected according to the sample product purchase record to obtain the original portrait structure data, then the weight assignment is performed on the portrait structure data according to the sample product purchase record, the portrait data structure is selected, then the product prediction is performed on the selected portrait data structure according to the preset product prediction model and the sample product purchase record to obtain the predicted product purchase record, and finally the parameter optimization is performed on the preset product prediction model according to the predicted product purchase record and the sample product purchase record to obtain the target product prediction model, thereby realizing obtaining a model capable of accurately predicting the user according to the user image, and further improving the recommendation success rate of the insurance product.
In step S301 of some embodiments, the sample user image is a graph data structure with a sample object master node and user information of the sample user as child nodes, each child node is connected to the master node, previous weights between the child nodes are calculated according to the selected edge weights, and on the basis, the sample product purchase record is abstracted into child nodes, and the child nodes are connected with the master node as a child node for the same insurance product, so as to obtain the original graph structure data. For example, the sample product purchase record describes that the sample object purchases the first product three times and the second product once, and then the first product serves as a child node and the second product serves as a child node, and is respectively connected with the main node formed by the sample object.
In step S302 of some embodiments, the image structure data is assigned with a weight according to the number of purchases of the product as the weight according to the sample product purchase record, for example, if the sample product purchase record records that the sample object purchases the first product three times, the edge weight between the main node formed by the sample object and the sub node formed by the first product is assigned to be 3.
In some embodiments, in step S303, the preset product prediction model is a model formed by a neural network, and in one embodiment, the preset product prediction model is NGCF (Neural Graph Coll aborative Filtering) models, and the selected image data structure, the sample product purchase record and the product information of the preset insurance product are input to NGCF to obtain the predicted product purchase record.
Referring to fig. 4, in some embodiments, step S304 may include, but is not limited to, steps S401 to S403:
step S401, performing sparse multi-category cross entropy loss calculation according to the predicted product purchase record and the sample product purchase record to obtain a sparse multi-category cross entropy loss value;
step S402, calculating focus loss according to the predicted product purchase record and the sample product purchase record to obtain a focus loss value;
And S403, carrying out parameter optimization on the preset product prediction model according to the focus loss value and the sparse multi-category cross entropy loss value to obtain a target product prediction model.
According to the steps S401 to S403 shown in the embodiment of the application, sparse multi-category cross entropy loss calculation is performed according to predicted product purchase records and sample product purchase records to obtain sparse multi-category cross entropy loss values, meanwhile, focus loss calculation is performed according to the predicted product purchase records and sample product purchase records to obtain focus loss values, finally, parameter optimization is performed on a preset product prediction model by the focus loss values and the sparse multi-category cross entropy loss values to obtain a target product prediction model, and a model capable of accurately predicting a user according to user images is obtained, so that the recommendation success rate of insurance products is improved.
In step S401 of some embodiments, the sparse multi-category cross entropy loss calculation is a calculation method for measuring the degree of difference between the predicted product purchase record and the sample product purchase record, and specifically is shown in formula (1):
Wherein N is the total number of sample objects, i is the index of the sample objects, C is the total number of preset insurance products, j is the index of the preset insurance products, yij is the sample product purchase record, and pij predicts the product purchase record.
In step S402 of some embodiments, the focus loss calculation is a calculation method for measuring the degree of difference between the predicted product purchase record and the sample product purchase record, specifically as shown in formula (2):
Wherein N is the total number of sample objects, i is the index of the sample objects, C is the total number of preset insurance products, j is the index of the preset insurance products, yij is the sample product purchase record, pij predicts the product purchase record, and γ is the preset focusing parameter.
In step S403 of some embodiments, a direct sum of the focus loss value and the sparse multi-class cross entropy loss value is obtained, and then a random gradient descent optimization is performed on the preset product prediction model according to the direct sum of the focus loss value and the sparse multi-class cross entropy loss value, so as to obtain a target product prediction model.
Referring to fig. 5, in some embodiments, step S104 includes, but is not limited to, steps S501 to S503:
step S501, carrying out purchasing power calculation according to the sample product purchasing record to obtain sample user purchasing power data;
Step S502, training a preset purchasing power prediction model according to the sample user portrait and sample user purchasing power data to obtain a target purchasing power prediction model;
and step S503, carrying out genetic search on the sample user portraits according to the target purchasing power prediction model to obtain the selected user portraits.
In the steps S501 to S503 shown in the embodiment of the present application, purchasing power calculation is performed according to a sample product purchasing record to obtain sample user purchasing power data, then a preset purchasing power prediction model is trained according to the sample user portrait and the sample user purchasing power data to obtain a target purchasing power prediction model, and finally genetic searching is performed on the sample user portrait according to the target purchasing power prediction model to obtain a selected user portrait, so that more data sets are obtained as much as possible, the data sets cover various user situations as much as possible, and genetic searching is performed on the sample user portrait based on the target purchasing power prediction model, so as to obtain the selected user portrait with high purchasing power.
In step S501 of some embodiments, the purchase power calculation is performed according to the sample product purchase record, which is to calculate the total value of the purchase products of the sample object on the preset insurance products, firstly obtain all the purchased insurance products and the number of times of purchase for the sample object, and then calculate the total value of the purchase products according to the number of times of purchase and the corresponding insurance products, so as to obtain the purchase power data of the sample user.
In step S502 of some embodiments, a sample user representation is input into a preset purchasing power prediction model, predicted purchasing power data is output by the preset purchasing power prediction model, then loss value calculation is performed based on the predicted purchasing power data and the sample user purchasing power data to obtain a training loss value, and then the preset purchasing power prediction model is trained based on the training loss value to obtain a target purchasing power prediction model. In one embodiment, the preset purchasing power prediction model is GAT (Graph Attention Networks) neural network model, the sample user portrait is input into the GAT to obtain predicted purchasing power data, mean square error calculation is performed based on the predicted purchasing power data and the sample user purchasing power data to obtain a training loss value, and then the preset purchasing power prediction model is trained based on the training loss value to obtain the target purchasing power prediction model.
Referring to fig. 6, in some embodiments, step S503 includes, but is not limited to, steps S601 to S606:
s601, carrying out chromosome coding on a sample user portrait to obtain an original population;
step S602, performing cross mutation on an original population to obtain a first population;
Step S603, screening the population fitness of the first population based on the target purchasing power prediction model to obtain a second population;
Step S604, carrying out local search on the second population based on the target purchasing power prediction model to obtain a third population;
step S605, filtering the population fitness of the first population based on the target purchasing power prediction model to obtain a fourth population;
and step S606, carrying out population combination based on the fourth population, the second population and the third population to obtain the selected user portrait.
In the steps S601 to S606 shown in the embodiment of the present application, chromosome encoding is performed on a sample user image to obtain an original population, then cross mutation is performed on the original population to obtain a first population, then population fitness screening is performed on the first population based on a target purchasing power prediction model to obtain a second population, then local searching is performed on the second population based on the target purchasing power prediction model to obtain a third population, population fitness filtering is performed on the first population based on the target purchasing power prediction model to obtain a fourth population, finally population merging is performed based on the fourth population, the second population and the third population to obtain a selected user image, so that more data sets are acquired as much as possible, the data sets cover various user situations as much as possible, and simultaneously, the user images stored in the data sets are screened through the target purchasing power prediction model, namely, the selected user image belongs to a user image with higher purchasing power, so as to provide a data basis for subsequent screening of a target object based on the selected user image.
In step S601 of some embodiments, the chromosome encoding converts the sample user portrait into an array, first extracts the child nodes in the sample user portrait as an array, and uses the values corresponding to the nodes as key value pairs of the array. For example, the child nodes of the sample user representation include age, income, occupation, and corresponding edges between the child nodes, i.e., age, income, are extracted, and the values of the child nodes are extracted to form key-value pairs, thereby forming an original population, e.g., the first sample object appears in the population as { first sample object: [ (age: 18), (income: 5000) ] }. The original population is a collection of multiple sample objects.
In step S602 of some embodiments, cross-mutating the original population is to select, according to each object in the original population, a node for exchange in a random manner, e.g., when the node for random exchange is an age, corresponding exchanges are performed for ages in the first sample object and the second sample object. And when the node which is randomly switched is income, carrying out corresponding switching on the income in the first sample object and the second sample object. The intersected population is then subjected to variation, i.e. the nodes of variation are selected in a randomly selected manner, e.g. to the age of the first sample object, and the age is modified according to a randomly selected value.
In step S603 of some embodiments, population fitness screening is performed on the first population based on the target purchasing power prediction model, wherein the first population is subjected to portrait construction, then portraits corresponding to the first population are input into the target purchasing power prediction model, predicted purchasing power data of the first population is output by the target purchasing power prediction model, and then the first population is screened according to a preset threshold screening and the predicted purchasing power data of the first population, so as to obtain the second population. For example, the first 10% of the first population is screened to obtain the second population.
In some embodiments, in step S604, the local search is performed on the second population based on the target purchasing power prediction model by taking the area where the second population is located as the center, screening the values near the center formed by the second population according to a preset threshold, for example, the second population comprises [ the first sample object (age: 19), (income: 12000) ], the preset age threshold is 1, and the income threshold is 1000, and four populations of [ (age: 20), (income: 12000), (age: 18), (income: 12000), (age: 19), (income: 13000), (age: 19), (income: 11000) ] can be obtained after the local search, then the image of the locally searched population is constructed, and the image is input into the target purchasing power prediction model, so that the chromosome with high purchasing power in the second population is obtained.
In step S605 of some embodiments, population fitness filtering is performed on the first population based on the target purchasing power prediction model, wherein population fitness filtering is performed on the first population, wherein the first population is subjected to image construction, then images corresponding to the first population are input into the target purchasing power prediction model, predicted purchasing power data of the first population is output by the target purchasing power prediction model, and then the first population is filtered according to preset filtering and predicted purchasing power data of the first population, so as to obtain the second population. For example, 10% of the first population is filtered out to obtain a fourth population.
In step S606 of some embodiments, the fourth population, the second population, and the third population are combined, i.e., the fourth population, the second population, and the third population are combined, and then the combined populations are subjected to image construction to obtain the selected user image.
Referring to fig. 7, in some embodiments, step S105 may include, but is not limited to, steps S701 through S703:
Step S701, constructing a user portrait according to candidate user information to obtain a candidate user portrait;
Step S702, performing cosine similarity screening according to the candidate user portraits and the selected user portraits to obtain target user portraits;
Step S703, screening the candidate object according to the target user portrait to obtain the target object.
In the steps S701 to S703 shown in the embodiment of the present application, a candidate user image is obtained by constructing a user image according to candidate user information, then cosine similarity screening is performed according to the candidate user image and a selected user image to obtain a target user image, and finally, a candidate object is screened according to the target user image to obtain a target object, thereby screening a target object with a potential high purchasing power from the candidate user information, and further improving the recommendation success rate of product recommendation for a subsequent process.
In step S701 of some embodiments, the principle of user portrayal construction based on candidate user information is similar to that of user portrayal construction based on sample user information, and will not be described here.
In some embodiments, in step S702, cosine similarity filtering is performed according to the candidate user image and the selected user portrait by first converting the image data structure of the candidate user image into a vector data structure, simultaneously converting the selected user portrait into a vector data structure, and then calculating cosine similarity between the vector data structure of the candidate user image and the vector data structure of the selected user portrait, and when the cosine similarity is smaller than a preset cosine threshold, taking the candidate user image as the target user portrait. In one embodiment, the user representation is converted into a vector data structure based on a pre-trained Graph2Vec model.
In step S703 of some embodiments, candidate objects are filtered according to the target user portrait, that is, the object corresponding to the target user portrait is obtained, that is, the target object is obtained.
In step S106 of some embodiments, a user representation of the target object is input into the target product prediction model to obtain a target product, and then the target product is used for recommending the target object, so that accurate screening of the user is achieved, and then the screened target user is accurately recommended by using the target product prediction model, so that the recommendation success rate of product recommendation is improved.
Referring to fig. 8, an embodiment of the present application further provides a product recommendation device, which may implement the product recommendation method, where the device includes:
an acquisition data module 801, configured to acquire sample user information of a sample object and a sample product purchase record of the sample object;
A portrait construction module 802, configured to perform user portrait construction according to sample user information, so as to obtain a sample user portrait;
the model training module 803 is configured to train the preset product prediction model according to the sample user portrait and the sample product purchase record, so as to obtain a target product prediction model;
A genetic search module 804, configured to perform genetic search on the sample product purchase record and the sample user representation to obtain a selected user representation;
the user screening module 805 is configured to obtain candidate user information of the candidate object, and perform similar user screening on the candidate object according to the selected user portrait and the candidate user information to obtain a target object;
The product recommendation module 806 is configured to predict a product of the target object based on the target product prediction model, obtain a target product, and recommend the target product to the target object.
The specific implementation of the product recommendation device is basically the same as the specific embodiment of the product recommendation method, and is not described herein.
The embodiment of the application also provides electronic equipment, which comprises a memory and a processor, wherein the memory stores a computer program, and the processor realizes the product recommendation method when executing the computer program. The electronic equipment can be any intelligent terminal including a tablet personal computer, a vehicle-mounted computer and the like.
Referring to fig. 9, fig. 9 illustrates a hardware structure of an electronic device according to another embodiment, the electronic device includes:
The processor 901 may be implemented by a general purpose CPU (central processing unit), a microprocessor, an application specific integrated circuit (ApplicationSpecificIntegratedCircuit, ASIC), or one or more integrated circuits, etc. for executing related programs, so as to implement the technical solution provided by the embodiments of the present application;
The memory 902 may be implemented in the form of read-only memory (ReadOnlyMemory, ROM), static storage, dynamic storage, or random access memory (RandomAccessMemory, RAM), among others. The memory 902 may store an operating system and other application programs, and when the technical solutions provided in the embodiments of the present disclosure are implemented by software or firmware, relevant program codes are stored in the memory 902, and the processor 901 invokes a product recommendation method for executing the embodiments of the present disclosure;
An input/output interface 903 for inputting and outputting information;
The communication interface 904 is configured to implement communication interaction between the device and other devices, and may implement communication in a wired manner (e.g. USB, network cable, etc.), or may implement communication in a wireless manner (e.g. mobile network, WIFI, bluetooth, etc.);
A bus 905 that transfers information between the various components of the device (e.g., the processor 901, the memory 902, the input/output interface 903, and the communication interface 904);
Wherein the processor 901, the memory 902, the input/output interface 903 and the communication interface 904 are communicatively coupled to each other within the device via a bus 905.
The embodiment of the application also provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program realizes the product recommendation method when being executed by a processor.
The memory, as a non-transitory computer readable storage medium, may be used to store non-transitory software programs as well as non-transitory computer executable programs. In addition, the memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory optionally includes memory remotely located relative to the processor, the remote memory being connectable to the processor 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 product recommending method, the product recommending device, the electronic equipment and the storage medium provided by the embodiment of the application are characterized in that the sample user information of a sample object and the sample product purchasing record of the sample object are obtained, then the user portrayal construction is carried out according to the sample user information to obtain the sample user portrayal, then a preset product predicting model is trained according to the sample user portrayal and the sample product purchasing record to obtain a model capable of accurately predicting a user according to the user image to improve the recommending success rate of an insurance product, further, the sample product purchasing record and the sample user portrayal are subjected to genetic search to obtain a selected user portrayal, so that more data sets are obtained as much as possible, so that the data sets cover various user conditions as much as possible, further, candidate user information of candidate objects is obtained, and similar user screening is carried out on the candidate objects according to the selected user information to obtain a target object, thus the recommending success rate of the insurance product is improved by finding out a user requiring product recommendation from a huge customer group, finally, the target product predicting model is predicted by accurately determining the user requiring product recommendation, the target product is predicted by the target product predicting model, the target product is obtained by carrying out on the target product predicting model, and the target product recommending is accurately screened by the recommended by the user on the recommended product and the user is screened according to the user requirements, the recommendation success rate of insurance products is improved.
The embodiments described in the embodiments of the present application are for more clearly describing the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application, and those skilled in the art can know that, with the evolution of technology and the appearance of new application scenarios, the technical solutions provided by the embodiments of the present application are equally applicable to similar technical problems.
It will be appreciated by persons skilled in the art that the embodiments of the application are not limited by the illustrations, and that more or fewer steps than those shown may be included, or certain steps may be combined, or different steps may be included.
The above described apparatus embodiments are merely illustrative, wherein the units illustrated as separate components may or may not be physically separate, i.e. may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
Those of ordinary skill in the art will appreciate that all or some of the steps of the methods, systems, functional modules/units in the devices disclosed above may be implemented as software, firmware, hardware, and suitable combinations thereof.
The terms "first," "second," "third," "fourth," and the like in the description of the application and in the above figures, if any, 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 the embodiments of the 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.
It should be understood that in the present application, "at least one (item)" means one or more, and "a plurality" means two or more. "and/or" is used to describe an association relationship of an associated object, and indicates that three relationships may exist, for example, "a and/or B" may indicate that only a exists, only B exists, and three cases of a and B exist simultaneously, where a and B may be singular or plural. The character "/" generally indicates that the context-dependent object is an "or" relationship. "at least one of" or the like means any combination of these items, including any combination of single item(s) or plural items(s). For example, at least one of a, b or c may represent a, b, c, "a and b", "a and c", "b and c", or "a and b and c", wherein a, b, c may be single or plural.
In the several embodiments provided by the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the above-described division of units is merely a logical function division, and there may be another division manner in actual implementation, for example, a plurality of units or components may be combined or may be 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 an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form.
The units described above as separate components may or may not be physically separate, and components 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 the embodiments 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 multiple 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 method of the various embodiments of the present application. The storage medium includes various media capable of storing programs, such as a U disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory RAM), a magnetic disk, or an optical disk.
The preferred embodiments of the present application have been described above with reference to the accompanying drawings, and are not thereby limiting the scope of the claims of the embodiments of the present application. Any modifications, equivalent substitutions and improvements made by those skilled in the art without departing from the scope and spirit of the embodiments of the present application shall fall within the scope of the claims of the embodiments of the present application.