Disclosure of Invention
The embodiment of the application aims to provide a data recommendation method, a data recommendation device, computer equipment and a storage medium, so as to solve the problem that accurate and personalized data recommendation cannot be performed on a user.
In order to solve the above technical problems, the embodiment of the present application provides a data recommendation method, which adopts the following technical scheme:
Acquiring commodity transaction logs, and extracting commodity metadata and user historical behavior data from the commodity transaction logs;
carrying out vectorization representation on the commodity metadata and the user historical behavior data to obtain commodity feature vectors and user feature vectors;
Performing similarity matching according to the commodity feature vector and the user feature vector to obtain a recommended commodity set;
extracting features of the recommended commodity set to obtain recommended commodity attribute features, and inputting the recommended commodity attribute features into a pre-trained commodity evaluation model to obtain a recommended commodity score;
Generating a recommended commodity list according to the recommended commodity score, and clustering and grouping the recommended commodity list to obtain a recommended commodity group;
and acquiring user real-time behavior data, adjusting the recommended commodity group according to the user real-time behavior data to obtain an effective recommended commodity group, and recommending data according to the effective recommended commodity group.
Further, the step of obtaining a commodity transaction log and extracting commodity metadata and user historical behavior data from the commodity transaction log specifically includes:
Acquiring a transaction log extraction identifier;
Extracting the commodity transaction log from a database according to the transaction log extraction identification;
and analyzing the commodity transaction log to obtain the commodity metadata and the user historical behavior data.
Further, the step of vectorizing the commodity metadata and the user historical behavior data to obtain a commodity feature vector and a user feature vector specifically includes:
Extracting features of the commodity metadata and the user historical behavior data to obtain commodity interaction features and user historical behavior features;
Performing first dimension reduction processing on the commodity interaction characteristics to obtain commodity characteristic vectors;
and performing second dimension reduction processing on the user history behavior characteristics to obtain the user characteristic vector.
Further, the step of obtaining a recommended commodity set by performing similarity matching according to the commodity feature vector and the user feature vector specifically includes:
cosine similarity calculation is carried out on the commodity feature vector based on a collaborative filtering algorithm to obtain commodity similarity, and a first recommended commodity is generated according to the commodity similarity;
Cosine similarity calculation is carried out on the user feature vector based on a collaborative filtering algorithm to obtain user similarity, and a second recommended commodity is generated according to the user similarity;
and integrating data of the first recommended commodity and the second recommended commodity to obtain the recommended commodity set.
Further, the step of extracting features of the recommended commodity set to obtain recommended commodity attribute features, and inputting the recommended commodity attribute features to a pre-trained commodity evaluation model to obtain a recommended commodity score specifically includes:
preprocessing the recommended commodity set to obtain a standard recommended commodity set;
acquiring commodity attribute information, and extracting features of the standard recommended commodity set according to the commodity attribute information to obtain the recommended commodity attribute features;
And obtaining a model extraction identifier, extracting the commodity evaluation model from a database according to the model extraction identifier, and inputting the attribute characteristics of the recommended commodity into the commodity evaluation model to obtain the score of the recommended commodity.
Further, the step of generating a recommended commodity list according to the recommended commodity score and clustering the recommended commodity list to obtain a recommended commodity group specifically includes:
Sorting the commodities according to the recommended commodity scores to obtain a recommended commodity sorting table;
Acquiring a recommendation score threshold, and screening the recommended commodity sorting table according to the recommendation score threshold to obtain the recommended commodity list;
Clustering and grouping the recommended commodity list according to a clustering algorithm to obtain a recommended commodity cluster;
Acquiring category information of the recommended commodity clusters, and acquiring a category ordering strategy according to the category information;
And sorting and screening the recommended commodity clusters according to the category sorting strategy to obtain the recommended commodity group.
Further, the step of obtaining user real-time behavior data, and adjusting the recommended commodity group according to the user real-time behavior data to obtain an effective recommended commodity group specifically includes:
Acquiring a real-time behavior data extraction identifier, and extracting the user real-time behavior data from a database according to the real-time behavior data extraction identifier;
performing behavior analysis on the user real-time behavior data to obtain user behavior information;
And carrying out commodity matching, priority sorting and group optimization processing on the recommended commodity group according to the user behavior information to obtain the effective recommended commodity group.
In order to solve the above technical problems, the embodiment of the present application further provides a data recommendation device, which adopts the following technical scheme:
The data acquisition module is used for acquiring commodity transaction logs, and extracting commodity metadata and user historical behavior data from the commodity transaction logs;
the data conversion module is used for vectorizing the commodity metadata and the user historical behavior data to obtain commodity feature vectors and user feature vectors;
The data matching module is used for carrying out similarity matching according to the commodity feature vector and the user feature vector to obtain a recommended commodity set;
The data evaluation module is used for extracting the characteristics of the recommended commodity set to obtain recommended commodity attribute characteristics, and inputting the recommended commodity attribute characteristics into a pre-trained commodity evaluation model to obtain a recommended commodity score;
the data classification module is used for generating a recommended commodity list according to the recommended commodity scores, and clustering and grouping the recommended commodity list to obtain a recommended commodity group;
The data recommendation module is used for acquiring real-time behavior data of a user, adjusting the recommended commodity group according to the real-time behavior data of the user to obtain an effective recommended commodity group, and recommending data according to the effective recommended commodity group.
In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical schemes:
A computer device comprising a memory having stored therein computer readable instructions which when executed by a processor implement the steps of the data recommendation method of any of the preceding claims.
In order to solve the above technical problems, an embodiment of the present application further provides a computer readable storage medium, which adopts the following technical schemes:
a computer readable storage medium having stored thereon computer readable instructions which when executed by a processor implement the steps of the data recommendation method according to any of the preceding claims.
Compared with the prior art, the method and the device have the advantages that commodity metadata and user historical behavior data are extracted from commodity transaction logs through obtaining the commodity transaction logs, vectorization representation is conducted on the commodity metadata and the user historical behavior data to obtain commodity feature vectors and user feature vectors, similarity matching is conducted on the commodity feature vectors and the user feature vectors to obtain a recommended commodity set, feature extraction is conducted on the recommended commodity set to obtain recommended commodity attribute features, the recommended commodity attribute features are input into a pre-trained commodity evaluation model to obtain recommended commodity scores, a recommended commodity list is generated according to the recommended commodity scores, clustering grouping is conducted on the recommended commodity list to obtain a recommended commodity set, real-time behavior data of a user are obtained, adjustment is conducted on the recommended commodity set according to the user real-time behavior data to obtain an effective recommended commodity set, and data recommendation is conducted according to the effective recommended commodity set. Therefore, accurate and personalized data recommendation is effectively performed on the user, and the conversion rate of the user is improved.
Detailed Description
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 terms used in the description herein are used for the purpose of describing particular embodiments only and are not intended to limit the application, and the terms "comprising" and "having" and any variations thereof in the description of the application and the claims and the above description of the drawings are intended to cover non-exclusive inclusions. The terms first, second and the like in the description and in the claims or in the above-described figures, are used for distinguishing between different objects and not necessarily for describing a sequential or chronological order.
Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are non-related or alternative embodiments mutually exclusive of other embodiments. Those of skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments.
In order to make the person skilled in the art better understand the solution of the present application, the technical solution of the embodiment of the present application will be clearly and completely described below with reference to the accompanying drawings.
As shown in fig. 1, a system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used as a medium to provide communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.
The user may interact with the server 105 via the network 104 using the terminal devices 101, 102, 103 to receive or send messages or the like. Various communication client applications, such as a web browser application, a shopping class application, a search class application, an instant messaging tool, a mailbox client, social platform software, etc., may be installed on the terminal devices 101, 102, 103.
The terminal devices 101, 102, 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smartphones, tablet computers, electronic book readers, MP3 players (Moving Picture ExpertsGroup Audio Layer III, dynamic video expert compression standard audio plane 3), MP4 (Moving PictureExperts Group Audio Layer IV, dynamic video expert compression standard audio plane 4) players, laptop and desktop computers, and the like.
The server 105 may be a server providing various services, such as a background server providing support for pages displayed on the terminal devices 101, 102, 103.
It should be noted that, the data recommendation method provided in the embodiment of the present application is generally executed by a server, and accordingly, the data recommendation device is generally disposed in the server.
It should be understood that the number of terminal devices, networks and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.
With continued reference to FIG. 2, a flow chart of one embodiment of a method of data recommendation in accordance with the present application is shown. The data recommendation method comprises the following steps:
Step S10, acquiring commodity transaction logs, and extracting commodity metadata and user historical behavior data from the commodity transaction logs;
In this embodiment, the commodity transaction log includes transaction basic information including transaction time, specific date and time of transaction, information of both sides of transaction including names, IDs or account numbers of sellers and buyers, commodity information including names, specifications, models, numbers, unit prices and the like of commodities, transaction amount including total amount of commodity, tax, freight and the like, and payment means including payment treasures, weChat payments, bank card payments and the like. The transaction process information comprises order numbers, order states, transaction mutual information, communication records between buyers and sellers, such as chatting content, negotiation results and the like, and complaints and feedback, such as complaints, evaluation or feedback information aiming at the transaction, wherein the order numbers uniquely identify each transaction, the order states comprise various stages of recording orders, such as payment, delivery, completion, cancellation and the like. Financial and settlement information, namely income and expenditure, namely detailed record of income and expenditure conditions in transactions, and settlement time, namely settlement time and mode of transaction money. The security and risk information comprises abnormal transaction warning, system or manually identified abnormal transaction behavior, and risk assessment, namely a transaction risk assessment result.
Step S20, vectorizing the commodity metadata and the user historical behavior data to obtain commodity feature vectors and user feature vectors;
In this embodiment, the commodity metadata refers to data describing a commodity, including commodity basic information such as commodity name, commodity number, brand, manufacturer, production place, etc., commodity description information including detailed description, specification parameters, functional characteristics, usage instructions, etc., price and promotion information including selling price, original price, discount information, promotion, etc., inventory and logistics information including inventory number, delivery place, expected arrival time, etc., and classification and label including classification, label, keyword, etc. to which the commodity belongs. The user history behavior data refers to behavior data generated by a user in the process of accessing and using the system platform, and comprises browsing behaviors such as a commodity list browsed by the user, page stay time, browsing paths and the like, searching behaviors such as search keywords input by the user, search time, search result click conditions and the like, collecting and purchasing behaviors such as commodities collected by the user and commodities added into a shopping cart, purchasing behaviors such as purchasing records, purchasing time, purchasing amount, purchasing quantity and the like of the user, evaluating and feedback behaviors such as evaluating, grading, comment content and feedback comments of the user on the commodities. The commodity feature vector and the user feature vector are feature vectors corresponding to commodity metadata and user historical behavior data, and can be obtained according to feature extraction and dimension reduction processing of the commodity metadata and the user historical behavior data.
Step S30, matching the similarity between the commodity feature vector and the user feature vector to obtain a recommended commodity set;
In this embodiment, the recommended commodity set is a data set generated by performing similarity matching based on a commodity feature vector and a user feature vector, and the recommended commodity set includes commodity data obtained by performing demand matching on the commodity feature vector and the user feature vector.
Step S40, extracting features of the recommended commodity set to obtain recommended commodity attribute features, and inputting the recommended commodity attribute features into a pre-trained commodity evaluation model to obtain recommended commodity scores;
In the embodiment, a pre-trained commodity evaluation model adopts a logistic regression model, and the training steps of the logistic regression model comprise the steps of obtaining sample commodity data, carrying out data cleaning, noise removal and outlier processing on the sample commodity data to obtain standard sample commodity data, carrying out feature extraction on the standard sample commodity data to obtain sample commodity features, setting a layer structure of the logistic regression model, initializing model parameters, training the logistic regression model by taking the sample commodity features as a training data set, calculating a loss function, and optimizing the logistic regression model by adopting a gradient descent method to obtain a commodity grading model.
S50, generating a recommended commodity list according to the recommended commodity scores, and clustering and grouping the recommended commodity list to obtain a recommended commodity group;
In this embodiment, the recommended commodity score is an evaluation score corresponding to a commodity in a recommended commodity set, the recommended commodity list is list information obtained by sorting and screening according to the recommended commodity score, the clustering grouping refers to clustering the recommended commodity list according to a preset clustering algorithm, in this embodiment, the preset clustering algorithm may adopt a K-Means algorithm, the K-Means algorithm is a widely used clustering algorithm, and the dataset is divided into K clusters in an iterative manner, so that data points in each cluster are as similar as possible, and data points among different clusters are as different as possible, and the commodity can be clustered according to features (such as price, sales volume, user evaluation, etc.) through the K-Means algorithm, thereby identifying a commodity group with similar attributes.
And step S60, acquiring user real-time behavior data, adjusting the recommended commodity group according to the user real-time behavior data to obtain an effective recommended commodity group, and recommending data according to the effective recommended commodity group.
In this embodiment, the effective recommended commodity group refers to commodity information that is output to a display page by the system for data recommendation display, and after the effective recommended commodity group is obtained, the effective recommended commodity group is sent to a browsing display page of a corresponding user by the system for data display, so that data recommendation of the user is completed.
The embodiment comprises the steps of obtaining commodity transaction logs, extracting commodity metadata and user historical behavior data from the commodity transaction logs, vectorizing the commodity metadata and the user historical behavior data to obtain commodity feature vectors and user feature vectors, matching similarity according to the commodity feature vectors and the user feature vectors to obtain a recommended commodity set, carrying out feature extraction on the recommended commodity set to obtain recommended commodity attribute features, inputting the recommended commodity attribute features into a pre-trained commodity evaluation model to obtain recommended commodity scores, generating a recommended commodity list according to the recommended commodity scores, clustering the recommended commodity list to obtain a recommended commodity group, obtaining user real-time behavior data, adjusting the recommended commodity group according to the user real-time behavior data to obtain an effective recommended commodity group, and carrying out data recommendation according to the effective recommended commodity group. Therefore, accurate and personalized data recommendation is effectively performed on the user, and the conversion rate of the user is improved.
With continued reference to fig. 3, in some alternative implementations of the present embodiment, step S10 includes the steps of:
Step S101, acquiring a transaction log extraction identifier;
in this embodiment, the transaction log extraction identifier is a unique identifier of the corresponding commodity transaction log, and the transaction log extraction identifier may be a log ID or a UUID (universally unique identification code).
Step S102, extracting the commodity transaction log from a database according to the transaction log extraction identification;
in this embodiment, the transaction log extraction identifier is used to perform matching query in the database, and the matched information is extracted, so as to obtain the commodity transaction log. In this embodiment, the commodity transaction log is generated by collecting and recording various information of the interaction between the user and the commodity by the system, and storing the information in the database for updating in real time.
Step S103, analyzing the commodity transaction log to obtain commodity metadata and the user historical behavior data;
in this embodiment, the commodity transaction log may be parsed and information extracted according to the first query information corresponding to the commodity metadata and the second query information corresponding to the user historical behavior data, so as to effectively obtain the commodity metadata and the user historical behavior data.
According to the embodiment, the commodity transaction log is extracted from the database according to the transaction log extraction identification, and is analyzed, so that commodity metadata and user historical behavior data in the commodity transaction log are effectively obtained, and subsequent vectorization representation processing is facilitated.
With continued reference to fig. 4, in some alternative implementations of the present embodiment, step S20 includes the steps of:
step S201, extracting features of the commodity metadata and the user historical behavior data to obtain commodity interaction features and user historical behavior features;
In this embodiment, before feature extraction is performed on the commodity metadata and the user historical behavior data, data cleaning, noise removal and outlier processing are further required on the commodity metadata and the user historical behavior data, so as to obtain effective commodity metadata and user historical behavior data. The commodity interaction features comprise commodity clicking features, commodity browsing features, commodity purchasing features, commodity collecting features and commodity evaluating features, and the user history behavior features comprise user history browsing features, user history purchasing features, user history searching features, user history stay features and user history evaluating features.
Step S202, performing first dimension reduction processing on the commodity interaction characteristics to obtain commodity characteristic vectors;
In this embodiment, the first dimension reduction process may employ Principal Component Analysis (PCA), where the Principal Component Analysis (PCA) includes data normalization, where Xnorm = (X- μ)/σ, where Xnorm is a data matrix subjected to the data normalization, X is an original matrix of commodity interaction features, μ is a mean vector of features, σ is a standard deviation vector of features (or diagonal elements of a covariance matrix), the commodity interaction features are normalized such that a mean value of each feature is 0 and a variance is 1, and a covariance matrix is calculated, where XTnorm is a matrix transpose of Xnorm, n is a number of samples, the covariance matrix measures correlation between features in the data set, each sample has p features, and the covariance matrix is a matrix of p×p for n samples, where each element Cij represents a covariance between an i-th feature and a j-th feature. The eigenvalues and eigenvectors of the covariance matrix are calculated as Svi＝λi·vi, where λi is the ith eigenvalue, vi is the corresponding eigenvector, the eigenvalues represent the degree of data change (i.e., variance) in the direction of the corresponding eigenvector, and the eigenvectors give the direction of these changes. And selecting a principal component, namely selecting feature vectors corresponding to the first k largest feature values according to the sizes of the feature values, wherein the feature vectors are the directions of the principal component. Usually, the number k of principal components is selected to be smaller than the number p of original features, so as to achieve the purpose of dimension reduction. And (3) data projection, namely Xpca＝Xnorm·[v1,v2,…,vk, wherein Xpca is a dimension-reduced data matrix, the column number of the dimension-reduced data matrix is k, and the original data is projected onto the selected principal component to obtain the dimension-reduced commodity feature vector.
And step S203, performing second dimension reduction processing on the user history behavior characteristics to obtain the user characteristic vector.
In this embodiment, the second dimension reduction process may also use Principal Component Analysis (PCA), and the effective dimension reduction process is performed on the user history behavior feature through the principal component analysis, so as to obtain a user feature vector.
The embodiment obtains commodity interaction characteristics and user historical behavior characteristics by extracting characteristics of the commodity metadata and the user historical behavior data, obtains commodity characteristic vectors by performing first dimension reduction processing on the commodity interaction characteristics, and obtains the user characteristic vectors by performing second dimension reduction processing on the user historical behavior characteristics. Therefore, feature extraction and dimension reduction processing are effectively carried out according to commodity metadata and user historical behavior data, and commodity feature vectors and user feature vectors are effectively obtained.
With continued reference to fig. 5, in some alternative implementations of the present embodiment, step S30 includes the steps of:
step S301, cosine similarity calculation is carried out on the commodity feature vector based on a collaborative filtering algorithm, commodity similarity is obtained, and a first recommended commodity is generated according to the commodity similarity;
In this embodiment, the collaborative filtering algorithm is a recommendation algorithm that completely depends on the behavior relationship between the user and the commodity, and the basic principle is that the similarity between the users or the similarity between the commodities is found by analyzing the behavior and the preference of the users, and then the recommended commodity is generated based on the similarity. In this embodiment, the cosine similarity is used to calculate the similarity between different commodity feature vectors, and the cosine similarity measures the proximity of two commodity feature vectors in the direction, regardless of their size, so as to help us find out which commodities are similar in the interaction mode of the user. After the commodity similarity is obtained, other commodities similar to the commodities that they like before are recommended to the current user based on the commodity similarity, and the first recommended commodity is generated by selecting the N commodities most similar to the target commodity.
Step S302, cosine similarity calculation is carried out on the user feature vector based on a collaborative filtering algorithm to obtain user similarity, and a second recommended commodity is generated according to the user similarity;
In this embodiment, the collaborative filtering algorithm is a recommendation algorithm that completely depends on the behavior relationship between the user and the commodity, and the basic principle is that the similarity between the users or the similarity between the commodities is found by analyzing the behavior and the preference of the users, and then the recommended commodity is generated based on the similarity. In the embodiment, after the user similarity is obtained, a part of users most similar to the target user is selected as a similar user set based on the user similarity, scoring of the target user on unscored commodities is predicted by combining the scoring condition of the similar users and the similarity weight, and a second recommended commodity is generated for the target user according to the predicted scoring.
Step S303, performing data integration on the first recommended commodity and the second recommended commodity to obtain the recommended commodity set.
In this embodiment, the recommended product set is generated by combining the first recommended product and the second recommended product.
According to the embodiment, the commodity feature vector is subjected to cosine similarity calculation based on a collaborative filtering algorithm to obtain commodity similarity, a first recommended commodity is generated according to the commodity similarity, the user feature vector is subjected to cosine similarity calculation based on the collaborative filtering algorithm to obtain user similarity, a second recommended commodity is generated according to the user similarity, and data integration is performed on the first recommended commodity and the second recommended commodity, so that a recommended commodity set which comprises commodities similar to a target commodity and also contains user favorite commodities similar to the target user is obtained effectively, and feature extraction processing is convenient to follow-up.
With continued reference to fig. 6, in some alternative implementations of the present embodiment, step S40 includes the steps of:
step S401, preprocessing the recommended commodity set to obtain a standard recommended commodity set;
In this embodiment, preprocessing includes deduplication, checking if duplicate products exist in the recommended product set, deduplication if so, data cleansing, removing products that lack critical information (such as prices, titles, picture links, etc.), or filling in such information (e.g., using default values or estimating based on average values of other similar products), format normalization, ensuring that all product data are in a consistent format, such as price format, date format, etc., screening, further screening the product set according to business needs or user preferences, such as screening according to price range, brands, categories, etc.
Step S402, acquiring commodity attribute information, and extracting features of the standard recommended commodity set according to the commodity attribute information to obtain the recommended commodity attribute features;
In this embodiment, the commodity attribute information includes various information such as text description, title, price, brand, user evaluation, sales, and score of the commodity, and the recommended commodity attribute features include recommended commodity price, recommended commodity brand awareness, recommended commodity user score average, recommended commodity good score, and the like.
Step S403, a model extraction identifier is obtained, the commodity evaluation model is extracted from a database according to the model extraction identifier, and the recommended commodity attribute characteristics are input into the commodity evaluation model to obtain the recommended commodity score.
In this embodiment, the model extraction identifier is unique identifier information corresponding to the commodity evaluation model, and matching query is performed in the database according to the model extraction identifier, so as to obtain the commodity evaluation model. In this embodiment, the commodity evaluation model adopts a pre-trained deep learning model, and the deep learning model may adopt a BERT model, where the BERT model is a pre-trained language representation model based on a transducer structure, and the recommended commodity score is obtained by extracting key information in text data such as commodity description and user evaluation, and generating corresponding representation vectors, so as to predict the commodity evaluation score.
The embodiment obtains a standard recommended commodity set by preprocessing the recommended commodity set, obtains commodity attribute information, performs feature extraction on the standard recommended commodity set according to the commodity attribute information to obtain the recommended commodity attribute feature, obtains a model extraction identifier, extracts the commodity evaluation model from a database according to the model extraction identifier, and inputs the recommended commodity attribute feature into the commodity evaluation model, thereby obtaining a recommended commodity score for effectively evaluating the recommended commodity set, and facilitating subsequent generation of a recommended commodity list.
With continued reference to fig. 7, in some alternative implementations of the present embodiment, step S50 includes the steps of:
Step S501, commodity sorting is carried out according to the recommended commodity scores to obtain a recommended commodity sorting table;
In this embodiment, the recommended commodity set is ranked from high to low according to the score of the recommended commodity score, so as to obtain a recommended commodity ranking table.
Step S502, a recommendation score threshold is obtained, and the recommended commodity ordering list is screened according to the recommendation score threshold to obtain the recommended commodity list;
In this embodiment, the recommendation score threshold is a preset screening threshold, the recommendation score threshold may be set according to an average value of recommendation commodity scores corresponding to the recommendation commodity sorting table, and the recommendation commodity list is obtained by comparing the recommendation commodity score corresponding to the recommendation commodity in the recommendation commodity sorting table with the recommendation score threshold to obtain a recommendation commodity with a recommendation commodity score greater than or equal to the recommendation score threshold.
Step S503, clustering and grouping the recommended commodity list according to a clustering algorithm to obtain a recommended commodity cluster;
in the embodiment, the clustering algorithm adopts a K-Means algorithm, and the commodities in the recommended commodity list are used as input of the K-Means algorithm, the K-Means algorithm is executed, and the clustering result output by the K-Means algorithm is used as the recommended commodity cluster, so that the recommended commodity list is effectively clustered and grouped.
Step S504, category information of the recommended commodity clusters is obtained, and a category ordering strategy is obtained according to the category information;
In this embodiment, the category information is a category label corresponding to a recommended commodity cluster, and may be defined by a commodity attribute. The class sorting strategy is a sorting screening method set according to business requirements or user preferences, can sort according to the popularity degree, the user interest matching degree, the commodity diversity and the like, and screens according to the preset commodity display quantity.
And step S505, sorting and screening the recommended commodity clusters according to the category sorting strategy to obtain the recommended commodity group.
In this embodiment, the recommended commodity clusters are sorted and screened by a category sorting strategy, and the sorted and screened recommended commodity clusters are used as component parts of the recommended commodity group, so that the recommended commodity group is finally obtained.
According to the embodiment, a recommended commodity sorting table is obtained by sorting commodities according to the recommended commodity scores, a recommended commodity sorting threshold is obtained, the recommended commodity sorting table is screened according to the recommended commodity sorting threshold to obtain a recommended commodity list, the recommended commodity list is clustered according to a clustering algorithm to obtain recommended commodity clusters, category information of the recommended commodity clusters is obtained, a category sorting strategy is obtained according to the category information, sorting screening is conducted on the recommended commodity clusters according to the category sorting strategy, and therefore a high-quality and highly-correlated and diversified recommended commodity group is effectively obtained, and subsequent adjustment processing based on the recommended commodity group is facilitated.
With continued reference to fig. 8, in some alternative implementations of the present embodiment, step S60 includes the steps of:
step S601, acquiring a real-time behavior data extraction identifier, and extracting the user real-time behavior data from a database according to the real-time behavior data extraction identifier;
in this embodiment, the real-time behavior data extraction identifier is a unique data identifier corresponding to the real-time behavior data of the user, and the real-time behavior data of the user is effectively extracted by performing matching query in the database through the real-time behavior data extraction identifier.
Step S602, performing behavior analysis on the user real-time behavior data to obtain user behavior information;
In the embodiment, the step of behavior analysis comprises the steps of behavior classification, behavior sequence analysis, preference extraction and extraction, wherein the behavior classification is used for classifying user behaviors according to types, such as browsing, clicking, shopping cart adding, purchasing and the like, the behavior sequence analysis is used for analyzing the sequence and frequency of the user behaviors and identifying the behavior mode and the interest point of the user, and the preference extraction is used for extracting key information such as commodity preference, price sensitivity, brand preference and the like of the user according to the behavior data of the user. The user behavior information comprises basic behavior information, namely, browsing behavior, namely, behavior of a user for browsing different pages, commodities, contents and the like on a website, wherein the browsing behavior comprises browsing duration, page stay time, click rate and the like, searching behavior, namely, behavior of the user for searching specific keywords or contents in a search engine or a platform and reflecting interests and requirements of the user, and clicking behavior, namely, behavior of the user for clicking elements such as links, buttons, advertisements and the like, is an important mode of interaction between the user and the page contents. The transaction behavior information comprises purchasing behavior of a user for purchasing goods or services on an electronic commerce platform, wherein the purchasing behavior comprises the type, quantity, amount and the like of the purchased goods, the payment behavior comprises the behavior of selecting a payment mode and completing a payment flow by the user, the payment habit and preference of the user are reflected, and the return behavior comprises the return behavior of unsatisfied goods after the user purchases and is an important index for evaluating the quality of the goods and the satisfaction degree of the user by enterprises. The interactive behavior information comprises comment behaviors of users on products, services or contents, including text comments, scoring and the like, reflecting satisfaction and feedback of the users, sharing behaviors of the users on sharing contents, commodities or information to friends or social media, contributing to expansion of brand influence, praying and collecting behaviors of the users on the contents or commodities, and indicating approval or potential purchase intention of the users on the contents.
And step S603, carrying out commodity matching, priority sorting and group optimization processing on the recommended commodity group according to the user behavior information to obtain the effective recommended commodity group.
In this embodiment, commodity matching refers to screening commodities which are most matched with a user from a recommended commodity group according to the behavior preference and interest points of the user, priority ranking refers to priority ranking of commodities in the recommended commodity group according to the weight of the user behavior (such as the weight of purchasing behavior is higher than that of browsing behavior) and the correlation of the commodities, and group optimization refers to grouping optimization processing of the commodities in consideration of the commodity diversity and complementarity of the recommended commodity group, so that each group can cover different demands and interest points of the user. And carrying out commodity matching, priority sorting and group optimization processing on the recommended commodity group according to the user behavior information, so that the finally displayed effective recommended commodity group is further optimized according to the real-time user behavior information, and the browsing satisfaction degree of the user is improved.
According to the embodiment, the real-time behavior data of the user is extracted from the database according to the real-time behavior data extraction identification, behavior analysis is carried out on the real-time behavior data of the user to obtain user behavior information, commodity matching, priority ordering and group optimization processing are carried out on the recommended commodity group according to the user behavior information, so that the effective recommended commodity group which is diversified and accords with the user behavior intention is obtained, browsing satisfaction of the user is improved, and further the conversion rate of the user is improved.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by computer readable instructions stored in a computer readable storage medium that, when executed, may comprise the steps of the embodiments of the methods described above. The storage medium may be a nonvolatile storage medium such as a magnetic disk, an optical disk, a Read-Only Memory (ROM), or a random access Memory (Random Access Memory, RAM).
It should be understood that, although the steps in the flowcharts of the figures are shown in order as indicated by the arrows, these steps are not necessarily performed in order as indicated by the arrows. The steps are not strictly limited in order and may be performed in other orders, unless explicitly stated herein. Moreover, at least some of the steps in the flowcharts of the figures may include a plurality of sub-steps or stages that are not necessarily performed at the same time, but may be performed at different times, the order of their execution not necessarily being sequential, but may be performed in turn or alternately with other steps or at least a portion of the other steps or stages.
With further reference to fig. 9, as an implementation of the method shown in fig. 1, the present application provides an embodiment of a data recommendation device, where an embodiment of the device corresponds to the embodiment of the method shown in fig. 1, and the device may be specifically applied to various electronic devices.
As shown in fig. 9, the data recommendation device 700 in this embodiment includes a data acquisition module 701, a data conversion module 702, a data matching module 703, a data evaluation module 704, a data classification module 705, and a data recommendation module 706. Wherein:
The data acquisition module 701 is configured to acquire a commodity transaction log, and extract commodity metadata and user historical behavior data from the commodity transaction log;
The data conversion module 702 is configured to perform vectorization representation on the commodity metadata and the user historical behavior data, so as to obtain a commodity feature vector and a user feature vector;
the data matching module 703 is configured to perform similarity matching according to the commodity feature vector and the user feature vector, so as to obtain a recommended commodity set;
the data evaluation module 704 is configured to perform feature extraction on the recommended commodity set to obtain recommended commodity attribute features, and input the recommended commodity attribute features to a pre-trained commodity evaluation model to obtain a recommended commodity score;
The data classification module 705 is configured to generate a recommended commodity list according to the recommended commodity score, and cluster and group the recommended commodity list to obtain a recommended commodity group;
the data recommendation module 706 is configured to obtain user real-time behavior data, adjust the recommended commodity group according to the user real-time behavior data, obtain an effective recommended commodity group, and perform data recommendation according to the effective recommended commodity group.
The data recommending device is capable of acquiring commodity transaction logs, extracting commodity metadata and user historical behavior data from the commodity transaction logs, vectorizing the commodity metadata and the user historical behavior data to obtain commodity feature vectors and user feature vectors, matching similarity according to the commodity feature vectors and the user feature vectors to obtain a recommended commodity set, carrying out feature extraction on the recommended commodity set to obtain recommended commodity attribute features, inputting the recommended commodity attribute features into a pre-trained commodity evaluation model to obtain recommended commodity scores, generating a recommended commodity list according to the recommended commodity scores, clustering the recommended commodity list to obtain a recommended commodity group, acquiring user real-time behavior data, adjusting the recommended commodity group according to the user real-time behavior data to obtain an effective recommended commodity group, and carrying out data recommendation according to the effective recommended commodity group. Therefore, accurate and personalized data recommendation is effectively performed on the user, and the conversion rate of the user is improved.
In order to solve the technical problems, the embodiment of the application also provides computer equipment. Referring specifically to fig. 10, fig. 10 is a basic structural block diagram of a computer device according to the present embodiment.
The computer device 8 comprises a memory 81, a processor 82, a network interface 83 communicatively connected to each other via a system bus. It should be noted that only computer device 8 having components 81-83 is shown in the figures, but it should be understood that not all of the illustrated components are required to be implemented and that more or fewer components may be implemented instead. It will be appreciated by those skilled in the art that the computer device herein is a device capable of automatically performing numerical calculation and/or information processing according to a preset or stored instruction, and its hardware includes, but is not limited to, a microprocessor, an Application SPECIFIC INTEGRATED Circuit (ASIC), a Programmable gate array (Field-Programmable GATE ARRAY, FPGA), a digital Processor (DIGITAL SIGNAL Processor, DSP), an embedded device, and the like.
The computer equipment can be a desktop computer, a notebook computer, a palm computer, a cloud server and other computing equipment. The computer equipment can perform man-machine interaction with a user through a keyboard, a mouse, a remote controller, a touch pad or voice control equipment and the like.
The memory 81 includes at least one type of readable storage medium including flash memory, hard disk, multimedia card, card memory (e.g., SD or DX memory, etc.), random Access Memory (RAM), static Random Access Memory (SRAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), programmable Read Only Memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the storage 81 may be an internal storage unit of the computer device 8, such as a hard disk or a memory of the computer device 8. In other embodiments, the memory 81 may also be an external storage device of the computer device 8, such as a plug-in hard disk, a smart memory card (SMART MEDIA CARD, SMC), a Secure Digital (SD) card, a flash memory card (FLASH CARD) or the like, which are provided on the computer device 8. Of course, the memory 81 may also comprise both an internal memory unit of the computer device 8 and an external memory device. In this embodiment, the memory 81 is typically used to store an operating system and various application software installed on the computer device 8, such as computer readable instructions of a data recommendation method. Further, the memory 81 may be used to temporarily store various types of data that have been output or are to be output.
The processor 82 may be a central processing unit (Central Processing Unit, CPU), controller, microcontroller, microprocessor, or other data processing chip in some embodiments. The processor 82 is typically used to control the overall operation of the computer device 8. In this embodiment, the processor 82 is configured to execute computer readable instructions stored in the memory 81 or process data, for example, execute computer readable instructions of the data recommendation method.
The network interface 83 may comprise a wireless network interface or a wired network interface, which network interface 83 is typically used to establish a communication connection between the computer device 8 and other electronic devices.
The embodiment can acquire commodity transaction logs by adopting the computer equipment, extract commodity metadata and user historical behavior data from the commodity transaction logs, vectorize the commodity metadata and the user historical behavior data to obtain commodity feature vectors and user feature vectors, match similarity according to the commodity feature vectors and the user feature vectors to obtain a recommended commodity set, extract features of the recommended commodity set to obtain recommended commodity attribute features, input the recommended commodity attribute features into a pre-trained commodity evaluation model to obtain recommended commodity scores, generate a recommended commodity list according to the recommended commodity scores, cluster the recommended commodity list to obtain a recommended commodity group, acquire user real-time behavior data, adjust the recommended commodity group according to the user real-time behavior data to obtain an effective recommended commodity group, and recommend data according to the effective recommended commodity group. Therefore, accurate and personalized data recommendation is effectively performed on the user, and the conversion rate of the user is improved.
The present application also provides another embodiment, namely, a computer-readable storage medium storing computer-readable instructions executable by at least one processor to cause the at least one processor to perform the steps of the data recommendation method as described above.
The method comprises the steps of obtaining a commodity transaction log, extracting commodity metadata and user historical behavior data from the commodity transaction log, vectorizing the commodity metadata and the user historical behavior data to obtain commodity feature vectors and user feature vectors, matching similarity according to the commodity feature vectors and the user feature vectors to obtain a recommended commodity set, extracting features of the recommended commodity set to obtain recommended commodity attribute features, inputting the recommended commodity attribute features into a pre-trained commodity evaluation model to obtain recommended commodity scores, generating a recommended commodity list according to the recommended commodity scores, clustering the recommended commodity list to obtain a recommended commodity set, obtaining user real-time behavior data, adjusting the recommended commodity set according to the user real-time behavior data to obtain an effective recommended commodity set, and recommending data according to the effective recommended commodity set. Therefore, accurate and personalized data recommendation is effectively performed on the user, and the conversion rate of the user is improved.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. 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 (e.g. ROM/RAM, magnetic disk, optical disk) comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the method according to the embodiments of the present application.
It is apparent that the above-described embodiments are only some embodiments of the present application, but not all embodiments, and the preferred embodiments of the present application are shown in the drawings, which do not limit the scope of the patent claims. This application may be embodied in many different forms, but rather, embodiments are provided in order to provide a thorough and complete understanding of the present disclosure. Although the application has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments described in the foregoing description, or equivalents may be substituted for elements thereof. All equivalent structures made by the content of the specification and the drawings of the application are directly or indirectly applied to other related technical fields, and are also within the scope of the application.
The non-native company software tools or components present in the embodiments of the present application are presented by way of example only and are not representative of actual use.