Information push method, device, equipment and storage medium

Abstract

The present invention relates to the field of computer network technology, and in particular to an information push method, device, equipment and storage medium. By acquiring business data and user data, and extracting features from the business data, business data features are obtained, and business items are generated; user category information corresponding to the user is extracted from the user data, and on this basis, a first association value between the business item and the user is calculated; user behavior information is extracted from the user data, and a user-related item set is generated; a collaborative filtering recommendation algorithm is called to calculate similarity to obtain a second association value; a comprehensive association value is calculated based on the first association value and the second association value, and the corresponding business items are sorted based on the comprehensive association value, and pushed according to a preset push method; thereby solving the problem of low accuracy of information push in the business information push scenario in the prior art.

Description

Technical Field

[0001] The present invention relates to the field of computer network technology, and in particular to an information push method, device, equipment and storage medium. Background Art

[0002] At present, existing information push solutions in business information push scenarios are usually implemented based on full-text retrieval plus basic NLP (Natural Language Processing). The recommendation of search result items is actually also based on full-text retrieval. This retrieval processing leads to a low accuracy rate of information push in business information push scenarios.

[0003] In summary, the prior art has the problem of low accuracy in information push in business information push scenarios. Summary of the invention

[0004] The main purpose of the present application is to provide an information push method, device, equipment and storage medium to solve the problem of low accuracy of information push in business information push scenarios in the prior art.

[0005] The first aspect of the present invention provides an information push method, which includes: acquiring business data and user data; performing feature extraction on the business data to obtain business data features, and generating business items based on the business data features; parsing the user data to obtain user category information corresponding to the user, determining the association weight corresponding to the user according to the user category information, and calculating a first association value between the business item and the user according to the association weight; extracting user behavior information corresponding to the user from the user data, matching associated items according to the user behavior information, and generating a user associated item set corresponding to the user based on the associated items; calling a collaborative filtering recommendation algorithm to calculate the similarity between the business item and each element in the user associated item set to obtain a second association value; calculating a comprehensive association value based on the first association value and the second association value according to a preset comprehensive association value calculation method, and sorting the corresponding business items based on the comprehensive association value, and pushing them according to a preset push method.

[0006] Optionally, in a first implementation method of the first aspect of the present invention, the feature extraction of the business data to obtain business data features, and the generation of business items based on the business data features, include: parsing the business data to obtain numerical business data and categorical business data; normalizing the numerical business data to obtain numerical business data features; one-hot encoding the categorical business data to obtain categorical business data features; converting the numerical business data features and the categorical business data features into corresponding feature vectors, and projecting the feature vectors to preset business dimensions to obtain features of each of the business dimensions; according to the business dimensions, counting the features of each of the business dimensions to obtain statistical feature values ​​corresponding to each of the business dimensions; encapsulating the business dimensions, the statistical feature values ​​corresponding to the business dimensions, and the corresponding business data in a preset format to obtain business items.

[0007] Optionally, in a second implementation method of the first aspect of the present invention, the user data is parsed to obtain user category information corresponding to the user, the association weight corresponding to the user is determined based on the user category information, and the first association value between the business item and the user is calculated based on the association weight, including: parsing the user data to obtain user category information corresponding to each of the users; based on the user category information, querying the association weight corresponding to each user in a preset association weight table, wherein the association weight includes the dimension weight corresponding to each of the business dimensions; parsing the business items to obtain the statistical characteristic value corresponding to each of the business dimensions; based on the association weight corresponding to each user, multiplying the statistical characteristic value with the corresponding dimension weight and accumulating them to obtain the first association value.

[0008] Optionally, in a third implementation method of the first aspect of the present invention, matching related items according to the user behavior information and generating a user related item set corresponding to the user based on the related items include: parsing the user behavior information to obtain the user's operation information on historical business items, wherein the operation information includes the related items corresponding to the operation behavior and the degree of association between the operation behavior and the related items; storing the related items and the degree of association according to the corresponding relationship to obtain the user related item set.

[0009] Optionally, in a fourth implementation method of the first aspect of the present invention, the comprehensive correlation value is calculated based on the first correlation value and the second correlation value according to a preset comprehensive correlation value calculation method, and the corresponding business items are sorted based on the comprehensive correlation value, and pushed according to a preset push method, including: according to a preset correlation value weight, multiplying the first correlation value and the second correlation value with the corresponding correlation value weight and accumulating them to obtain a comprehensive correlation value; for each of the users, extracting the comprehensive correlation values ​​corresponding to all the business items, and sorting them to obtain a matter push sequence corresponding to each of the users; according to a preset push quantity threshold, selecting a corresponding number of business items from the matter push sequence, and pushing them to the corresponding users.

[0010] Optionally, in a fifth implementation of the first aspect of the present invention, after calculating a comprehensive correlation value based on the first correlation value and the second correlation value according to a preset comprehensive correlation value calculation method, sorting the corresponding business items based on the comprehensive correlation value, and pushing them according to a preset push method, it also includes: recording the user's operation sequence for the pushed business items to obtain user behavior information; obtaining the first correlation value and the second correlation value and the correlation value weight of the business item corresponding to the user behavior information to obtain correlation value information; calculating the correlation between the user behavior information and the correlation value information through a preset neural network model to obtain the correlation; and adjusting the correlation value weight according to the correlation.

[0011] Optionally, in a sixth implementation method of the first aspect of the present invention, after calling the collaborative filtering recommendation algorithm, calculating the similarity between the business items and the elements in the user-associated item set, and obtaining the second association value, it also includes: for each of the users, extracting the first association value and the second association value corresponding to all the business items, respectively, and sorting them to obtain the first push sequence and the second push sequence corresponding to each of the users; selecting a corresponding number of business items from the first push sequence and the second push sequence according to the preset first quantity threshold and the second quantity threshold, respectively; and pushing the corresponding number of business items to the corresponding user according to the preset push method.

[0012] The second aspect of the present invention provides an information push device, including: an acquisition module, which is used to acquire business data and user data; a first generation module, which is used to extract features from the business data, obtain business data features, and generate business items based on the business data features; a first calculation module, which is used to parse the user data, obtain user category information corresponding to the user, determine the association weight corresponding to the user according to the user category information, and calculate the first association value between the business item and the user according to the association weight; a second generation module, which is used to extract user behavior information corresponding to the user from the user data, match associated items according to the user behavior information, and generate a user associated item set corresponding to the user based on the associated items; a second calculation module, which is used to call a collaborative filtering recommendation algorithm, calculate the similarity between the business item and each element in the user associated item set, and obtain a second association value; a first push module, which is used to calculate a comprehensive association value based on the first association value and the second association value according to a preset comprehensive association value calculation method, sort the corresponding business items based on the comprehensive association value, and push them according to a preset push method.

[0013] Optionally, in a first implementation method of the second aspect of the present invention, the first generation module includes: a first parsing unit, used to parse the business data to obtain numerical business data and categorical business data; a normalization unit, used to normalize the numerical business data to obtain numerical business data features; an encoding unit, used to one-hot encode the categorical business data to obtain categorical business data features; a feature conversion unit, used to convert the numerical business data features and the categorical business data features into corresponding feature vectors, and project the feature vectors to preset business dimensions to obtain features of each of the business dimensions; a feature statistics unit, used to count the features of each of the business dimensions according to the business dimensions to obtain statistical feature values ​​corresponding to each of the business dimensions; a first generation unit, used to encapsulate the business dimensions, the statistical feature values ​​corresponding to the business dimensions and the corresponding business data in a preset format to obtain business matters.

[0014] Optionally, in a second implementation method of the second aspect of the present invention, the first calculation module includes: a second parsing unit, used to parse the user data to obtain user category information corresponding to each of the users; a determination unit, used to query the association weights corresponding to each of the users in a preset association weight table based on the user category information, wherein the association weights include dimension weights corresponding to each of the business dimensions; a third parsing unit, used to parse the business matters to obtain statistical characteristic values ​​corresponding to each of the business dimensions; and a first calculation unit, used to multiply and accumulate the statistical characteristic values ​​with the corresponding dimension weights based on the association weights corresponding to each of the users to obtain a first association value.

[0015] Optionally, in a third implementation of the second aspect of the present invention, the second generation module includes: a fourth parsing unit, used to parse the user behavior information to obtain the user's operation information on historical business items, wherein the operation information includes related items corresponding to the operation behavior and the degree of correlation between the operation behavior and the related items; a second generation unit, used to store the related items and the degree of correlation according to the corresponding relationship to obtain the user related item set.

[0016] Optionally, in a fourth implementation method of the second aspect of the present invention, the first push module includes: a second calculation unit, used to multiply and accumulate the first association value and the second association value with the corresponding association value weights according to a preset association value weight to obtain a comprehensive association value; a sorting unit, used to extract the comprehensive association values ​​corresponding to all business items for each of the users, and sort them to obtain a matter push sequence corresponding to each of the users; a push unit, used to select a corresponding number of business items from the matter push sequence according to a preset push quantity threshold, and push them to the corresponding users.

[0017] Optionally, in a fifth implementation of the second aspect of the present invention, the information push device also includes an adjustment module, which is used to record the user's operation sequence for the pushed business items to obtain user behavior information; obtain the first association value and the second association value and the association value weight of the business item corresponding to the user behavior information to obtain the association value information; calculate the correlation between the user behavior information and the association value information through a preset neural network model to obtain the correlation; and adjust the association value weight according to the correlation.

[0018] Optionally, in a sixth implementation method of the second aspect of the present invention, the information push device also includes a second push module, which is used to extract the first association value and the second association value corresponding to all business items for each of the users, and sort them to obtain the first push sequence and the second push sequence corresponding to each of the users; select a corresponding number of business items from the first push sequence and the second push sequence according to the preset first quantity threshold and the second quantity threshold, respectively; and push the corresponding number of business items to the corresponding user according to the preset push method.

[0019] The third aspect of the present invention provides a computer device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the computer device executes each step of the above-mentioned information push method.

[0020] A fourth aspect of the present invention provides a computer-readable storage medium, in which instructions are stored. When the computer-readable storage medium is run on a computer, the computer executes each step of the above-mentioned information push method.

[0021] In the technical solution of the present invention, the method specifically obtains business data and user data, and extracts features of the business data to obtain business data features and generate business items; extracts user category information corresponding to the user from the user data, and calculates a first association value between the business item and the user on this basis; extracts user behavior information from the user data and generates a set of user-related items; calls a collaborative filtering recommendation algorithm to calculate similarity and obtains a second association value; calculates a comprehensive association value based on the first association value and the second association value, and sorts the corresponding business items based on the comprehensive association value, and pushes them according to a preset push method; thereby solving the problem of low accuracy of information push in the business information push scenario in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a schematic diagram of a first embodiment of an information push method in an embodiment of the present invention;

[0023] Figure 2 Schematic diagram of a second embodiment of the information push method in an embodiment of the present invention;

[0024] Figure 3 Schematic diagram of a third embodiment of the information push method in an embodiment of the present invention;

[0025] Figure 4 A schematic diagram of an embodiment of an information push device in an embodiment of the present invention;

[0026] Figure 5 It is a schematic diagram of another embodiment of the information push device in the embodiment of the present invention;

[0027] Figure 6 FIG. 1 is a schematic diagram of an embodiment of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0028] In order to solve the problem of low accuracy of information push in business information push scenario in the prior art, the present application provides an information push method, device, equipment and storage medium. The method obtains business data and user data, extracts features of the business data, obtains business data features, and generates business items; extracts user category information corresponding to the user from the user data, and calculates the first association value between the business item and the user on this basis; extracts user behavior information from the user data, and generates a user-related item set; calls a collaborative filtering recommendation algorithm to calculate similarity and obtain a second association value; calculates a comprehensive association value based on the first association value and the second association value, and sorts the corresponding business items based on the comprehensive association value, and pushes according to a preset push method; thereby solving the problem of low accuracy of information push in business information push scenario in the prior art.

[0029] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0030] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , the first embodiment of the information push method in the embodiment of the present invention, the implementation steps of the method are as follows:

[0031] 101. Obtain business data and user data;

[0032] In this step, the business data at least includes data of the logistics business platform, and the user data at least includes data of the logistics management platform.

[0033] In actual applications, this step also includes transmitting the business data and user data to a corresponding backend database, wherein the backend database includes at least Greenplum, Mysql and Kudu databases.

[0034] 102. Extract features from the business data to obtain business data features, and generate business items based on the business data features;

[0035] This step can be implemented in the following ways:

[0036] Classify the business data according to data type to obtain numerical business data and categorical business data;

[0037] Performing numerical standardization processing on the numerical business data, and performing statistics in each preset numerical dimension to obtain numerical features, wherein the preset numerical dimensions at least include a customer increment dimension, a customer compliance dimension, and an average service quality score dimension;

[0038] Normalizing the numerical features to obtain numerical business data features;

[0039] Extracting features from the categorical business data and converting them into feature codes through One-Hot Encoding to obtain categorical business data features;

[0040] The numerical business data features and the categorical business data features are converted into corresponding feature vectors through a convolutional neural network, and the feature vectors are projected to preset business dimensions to obtain features of each of the business dimensions, wherein the business dimensions at least include a service evaluation dimension, a customer volume dimension, and a profit dimension;

[0041] According to the business dimensions, the features of each business dimension are scored to obtain a feature score value corresponding to each business dimension, for example, the corresponding feature score is obtained by querying a preset feature score table with the feature of the business dimension as an index;

[0042] The business dimension, the characteristic score value corresponding to the business dimension and the corresponding business data are packaged according to a preset format to obtain a business item.

[0043] 103. Parse the user data to obtain user category information corresponding to the user, determine the association weight corresponding to the user according to the user category information, and calculate the first association value between the business item and the user according to the association weight;

[0044] This step can be implemented in the following ways:

[0045] Parsing the user data to obtain user category information corresponding to each of the users;

[0046] Searching for a corresponding user tag according to the user category information;

[0047] According to the user tag, query the association weight corresponding to each user tag in a preset association weight table, wherein the association weight includes the dimension weight corresponding to each business dimension;

[0048] Analyze the business items to obtain characteristic score values ​​corresponding to each business dimension;

[0049] The feature score value is multiplied by the corresponding dimension weight and then added to obtain a first correlation value.

[0050] 104. Extracting user behavior information corresponding to the user from the user data, matching related items according to the user behavior information, and generating a user related item set corresponding to the user based on the related items;

[0051] This step can be implemented in the following ways:

[0052] Parsing the user data to obtain a user behavior sequence;

[0053] Parsing the user behavior sequence to obtain items associated with the user behavior, for example, if the user behavior sequence includes a click behavior on item A and a collection behavior on item B, then the items associated with the user behavior include item A and item B;

[0054] According to the preset correlation calculation method, the corresponding correlation is calculated according to the items associated with the user behavior. For example, if the user behavior sequence includes a click behavior on item A and a collection behavior on item B, the correlation corresponding to item A is 1, and the correlation corresponding to item B is 2;

[0055] The items associated with the user behavior and the corresponding association degrees are stored according to the corresponding relationship to obtain the user-associated item set.

[0056] 105. Calling a collaborative filtering recommendation algorithm to calculate the similarity between the business items and the elements in the user-related items set to obtain a second correlation value;

[0057] In this step, the process of calculating the similarity between the business items and the elements in the user-related items set includes calculating the similarity using the following formula:

[0058]

[0059] Among them, W ijrepresents the familiarity between item i and item j, N(i) is the number of users including item i in the user-related item set, N(j) is the number of users including item j in the user-related item set, and |N(i)∩N(j)| is the number of users including both item i and item j in the user-related item set.

[0060] In practical applications, the process of calculating the similarity between the business items and the elements in the user-related items set also includes:

[0061] The associations of users including item i in the user-associated item set and users including item j in the user-associated item set to the corresponding items are counted respectively, and the associations to the corresponding items are used as weights and substituted into N(i) and N(j) in the above formula respectively.

[0062] 106. According to a preset comprehensive correlation value calculation method, a comprehensive correlation value is calculated based on the first correlation value and the second correlation value, and corresponding business items are sorted based on the comprehensive correlation value, and pushed according to a preset push method.

[0063] This step can be implemented in the following ways:

[0064] According to the preset association value weight, the first association value and the second association value are multiplied by the corresponding association value weight and added to obtain a comprehensive association value. For example, when the association value weight is configured as "the first association value corresponding weight is 0.5 and the second association value corresponding weight is 0.5", the first association value and the second association value are multiplied by 0.5 respectively and added to obtain a comprehensive association value;

[0065] For each of the users, the comprehensive correlation values ​​corresponding to all business items are extracted and sorted to obtain the item push sequence corresponding to each of the users;

[0066] According to the preset push quantity threshold, a corresponding number of business items are selected from the item push sequence and pushed to the corresponding users.

[0067] In practical applications, this step also includes:

[0068] Record the user's operation behavior on the pushed business items and obtain the user operation sequence;

[0069] Parse the user operation sequence to obtain user behavior information, for example, record the click-to-open operation and the click-to-close operation for the same item in the user operation sequence, calculate the user's item processing time, and save it in a preset user behavior information table;

[0070] Obtain the business items corresponding to the user behavior information and obtain the behavior-related items;

[0071] Obtaining a first association value, a second association value and an association value weight corresponding to the behavior-related item to obtain association value information;

[0072] Calculating the correlation between the user behavior information and the associated value information through a preset neural network model to obtain a correlation degree, for example, calculating the correlation through a stacked autoencoder;

[0073] The association value weight is adjusted according to the relevance, for example, by inputting the user behavior information and the association value information into a multi-layer perceptron training to obtain a modified association value weight.

[0074] By implementing the above method, business data and user data are obtained, and feature extraction is performed on the business data to obtain business data features and generate business items; user category information corresponding to the user is extracted from the user data, and on this basis, a first association value between the business item and the user is calculated; user behavior information is extracted from the user data, and a user-related item set is generated; a collaborative filtering recommendation algorithm is called to calculate similarity to obtain a second association value; a comprehensive association value is calculated based on the first association value and the second association value, and the corresponding business items are sorted based on the comprehensive association value, and pushed according to a preset push method; In the above, business items are generated by extracting features of business data, and business information is encapsulated into business items with higher accuracy; the association value is calculated by combining the characteristics of each dimension of the business item through the preset association value weight to obtain the first association value, and the similarity between the business item and the association value is calculated based on the item set associated with the user behavior through the collaborative filtering recommendation algorithm to obtain the second association value, and the first association value and the second association value are combined according to the association value weight to push the information, which can effectively improve the accuracy of business information push, thereby solving the problem of low accuracy of information push in the business information push scenario in the prior art.

[0075] See also Figure 2 , the second embodiment of the information push method in the embodiment of the present invention, the implementation steps of the method are as follows:

[0076] 201. Obtain user data and business data;

[0077] This step can be implemented in the following ways:

[0078] Obtain business data from the business server;

[0079] The user data is obtained from a user database, wherein the user database includes corresponding Greenplum, Mysql and Kudu databases.

[0080] In practical applications, this step also includes:

[0081] The business data and the user data are transmitted to the Xingchen headquarters library by direct push or interface data transmission, wherein the Xingchen headquarters library is a database collection.

[0082] 202. Parse the user data to obtain user category information corresponding to the user, determine the data role corresponding to each user based on the user category information, and generate a corresponding user tag;

[0083] This step can be implemented in the following ways:

[0084] Parsing the user data to obtain a user identifier;

[0085] Using the user ID as an index, query the db_gxt.sys_user_role table to obtain the user category information corresponding to the user;

[0086] According to the user category information, by associating the report.rep_role_district table, the data role corresponding to the user is obtained;

[0087] A corresponding user tag is generated based on the data role.

[0088] 203. Count the business data according to the preset statistical dimensions to obtain statistical information, and generate business items based on the statistical information and the corresponding business data;

[0089] In this step, the statistical dimensions include at least: the dimension of business data that did not meet the standards on the day, the dimension of business decline on a month-on-month basis, the dimension of service quality, and the dimension of profit.

[0090] 204. Determine the association weight according to the user tag, and calculate the fixed push association value of the business item based on the association weight;

[0091] This step can be implemented in the following ways:

[0092] Using the user tag as an index, query the corresponding association weight in a preset association weight table;

[0093] Analyze the business items to obtain statistical information corresponding to each statistical dimension;

[0094] Normalizing the statistical information to obtain characteristic values ​​of each statistical dimension;

[0095] The characteristic values ​​of each statistical dimension are multiplied by the corresponding association weights, and the results are accumulated to obtain a fixed push association value.

[0096] 205. Analyze the user data to obtain user behavior characteristics, and calculate the recommendation association value based on the user behavior characteristics and business items;

[0097] This step can be implemented in the following ways:

[0098] Extract user behavior data from user data, and perform feature extraction through convolutional neural networks to obtain user behavior features;

[0099] Extracting the invisible relationship vector between the user behavior characteristics and business items by the stacked autoencoder;

[0100] Calculating the relevance between the user and the business item according to the invisible relationship vector;

[0101] Normalize the correlation to obtain the recommended correlation value

[0102] 206. Sort and push corresponding business items according to the fixed push association value and the recommended association value respectively.

[0103] This step can be implemented in the following ways:

[0104] For each of the users, respectively extract the fixed push association values ​​and the recommended association values ​​corresponding to all business items, and sort them to obtain the fixed push sequence and the recommended push sequence corresponding to each of the users;

[0105] According to the preset fixed position push threshold and the recommended position push threshold, a corresponding number of business items are selected from the fixed push sequence and the recommended push sequence respectively;

[0106] According to the preset push method, the corresponding number of business items are pushed to the corresponding users. For example, the business items selected from the fixed push sequence are pushed to the preset fixed position, and the business items selected from the recommended push sequence are pushed to the preset recommended position.

[0107] In actual applications, this step also includes determining the recommended position according to the following priorities:

[0108] Newly launched topics > Topics that did not reach monthly targets > Topics with more than 3 clicks yesterday > Topics that did not reach daily targets > Topics with lower rankings > Topics with a 10% month-on-month drop > Other topics.

[0109] By implementing the above method, business data and user data are obtained, and features are extracted from the business data to obtain business data features and generate business items; user category information corresponding to the user is extracted from the user data, and on this basis, a first association value between the business item and the user is calculated; user behavior information is extracted from the user data, and a user-related item set is generated; a collaborative filtering recommendation algorithm is called to calculate similarity to obtain a second association value; a comprehensive association value is calculated based on the first association value and the second association value, and the corresponding business items are sorted based on the comprehensive association value, and pushed according to a preset push method; the above method is implemented by obtaining user data and business data; extracting features from the user data to obtain business data features and generating business items; extracting user category information corresponding to the user from the user data, and on this basis, a first association value between the business item and the user is calculated; extracting user behavior information from the user data, and generating a user-related item set; calling a collaborative filtering recommendation algorithm to calculate similarity to obtain a second association value; calculating a comprehensive association value based on the first association value and the second association value, and sorting the corresponding business items based on the comprehensive association value, and pushing them according to a preset push method; Analyze and obtain user category information corresponding to the user, and determine the data role corresponding to each user based on the user category information, and generate corresponding user tags; perform statistics on business data according to preset statistical dimensions to obtain statistical information, and generate business items based on the statistical information and the corresponding business data; determine association weights according to user tags, and calculate fixed push association values ​​of business items based on the association weights; analyze user data to obtain user behavior characteristics, and calculate recommended association values ​​based on user behavior characteristics and business items; sort and push corresponding business items according to the fixed push association values ​​and recommended association values, respectively; thereby solving the problem of low accuracy of information push in business information push scenarios in the prior art.

[0110] See also Figure 3 , the third embodiment of the information push method in the embodiment of the present invention, the implementation steps of the method are as follows:

[0111] 301. Read data from the preset server;

[0112] In this step, the data at least includes user quantity label data, user benefit label data and user service quality label data;

[0113] The user volume label data includes information corresponding to users who did not meet the standard yesterday, users who ranked last, users who did not meet the standard today, and users with a large month-on-month decline;

[0114] The user profit tag data includes regular quantity tags and data corresponding to the theme of daily profit clearing;

[0115] The user service quality label data at least includes rookie ranking information.

[0116] 302. Extract user data from the data, and construct a user behavior sequence based on the user data;

[0117] This step can be implemented in the following ways:

[0118] Extracting features from the user data to obtain a user data feature vector;

[0119] Reducing the dimension of the user data feature vector by feature embedding to obtain a dense vector;

[0120] Obtaining user behavior features by embedding the dense vector;

[0121] A user behavior sequence is constructed based on the user behavior characteristics.

[0122] 303. Extract business data from the data, perform feature extraction, and obtain business dimension features;

[0123] This step can be implemented in the following ways:

[0124] Extracting features from the business data to obtain a business data feature vector;

[0125] Reduce the dimension of the business data feature vector by feature embedding to obtain a business dimension vector;

[0126] The business dimension feature is obtained by embedding the business dimension vector.

[0127] 304. Construct a sequence of information to be pushed based on the user behavior sequence and business dimension characteristics;

[0128] This step can be implemented in the following ways:

[0129] Calling a collaborative filtering recommendation algorithm to convert the user behavior sequence and business dimension features into a vector representation based on users and objects to obtain a vector matrix;

[0130] Calling a cluster analysis algorithm to perform cluster analysis based on the vector matrix to obtain cluster groups and corresponding recommended values;

[0131] The recommended values ​​are sorted to obtain a sequence of information to be pushed.

[0132] 305. Push information based on the sequence of information to be pushed.

[0133] This step can be implemented in the following ways:

[0134] According to the preset recommendation positions, a corresponding number of information to be pushed are selected from the sequence of information to be pushed;

[0135] The information to be pushed is pushed through a preset push method, for example, the information to be pushed is transmitted to the push interface through direct push or interface drop method.

[0136] By implementing the above method, business data and user data are obtained, and feature extraction is performed on the business data to obtain business data features and generate business items; user category information corresponding to the user is extracted from the user data, and on this basis, a first association value between the business item and the user is calculated; user behavior information is extracted from the user data, and a set of user-related items is generated; a collaborative filtering recommendation algorithm is called to calculate the similarity to obtain a second association value; a comprehensive association value is calculated based on the first association value and the second association value, and the corresponding business items are sorted based on the comprehensive association value, and pushed according to a preset push method; in the above, data is obtained from a preset server; user data is extracted from the data, and a user behavior sequence is constructed based on the user data; business data is extracted from the data, and feature extraction is performed to obtain business dimension features; a sequence of information to be pushed is constructed based on the user behavior sequence and the business dimension features; information is pushed based on the sequence of information to be pushed; thereby solving the problem of low accuracy of information push in the business information push scenario in the prior art.

[0137] The above describes the information push method in the embodiment of the present invention. The following describes the information push device in the embodiment of the present invention. Figure 4 , an embodiment of the information push device in the embodiment of the present invention, the device includes:

[0138] Acquisition module 401, used to acquire business data and user data;

[0139] A first generating module 402 is used to extract features from the business data to obtain business data features, and generate business items based on the business data features;

[0140] A first calculation module 403 is used to parse the user data to obtain user category information corresponding to the user, determine the association weight corresponding to the user according to the user category information, and calculate the first association value between the business item and the user according to the association weight;

[0141] A second generating module 404 is used to extract user behavior information corresponding to the user from the user data, match related items according to the user behavior information, and generate a user related item set corresponding to the user based on the related items;

[0142] The second calculation module 405 is used to call the collaborative filtering recommendation algorithm to calculate the similarity between the business items and the elements in the user-related items set to obtain a second correlation value;

[0143] The first push module 406 is used to calculate a comprehensive correlation value based on the first correlation value and the second correlation value according to a preset comprehensive correlation value calculation method, sort the corresponding business items based on the comprehensive correlation value, and push them according to the preset push method.

[0144] Through the implementation of the above-mentioned device, by acquiring business data and user data, and extracting features from the business data, business data features are obtained, and business items are generated; user category information corresponding to the user is extracted from the user data, and on this basis, a first association value between the business item and the user is calculated; user behavior information is extracted from the user data, and a set of user-related items is generated; a collaborative filtering recommendation algorithm is called to calculate similarity to obtain a second association value; a comprehensive association value is calculated based on the first association value and the second association value, and the corresponding business items are sorted based on the comprehensive association value, and pushed according to a preset push method; thereby solving the problem of low accuracy of information push in the business information push scenario in the prior art.

[0145] See also Figure 5 Another embodiment of the information push device in the embodiment of the present invention includes:

[0146] Acquisition module 401, used to acquire business data and user data;

[0147] A first generating module 402 is used to extract features from the business data to obtain business data features, and generate business items based on the business data features;

[0148] A first calculation module 403 is used to parse the user data to obtain user category information corresponding to the user, determine the association weight corresponding to the user according to the user category information, and calculate the first association value between the business item and the user according to the association weight;

[0149] A second generating module 404 is used to extract user behavior information corresponding to the user from the user data, match related items according to the user behavior information, and generate a user related item set corresponding to the user based on the related items;

[0150] The second calculation module 405 is used to call the collaborative filtering recommendation algorithm to calculate the similarity between the business items and the elements in the user-related items set to obtain a second correlation value;

[0151] A first push module 406, configured to calculate a comprehensive relevance value based on the first relevance value and the second relevance value according to a preset comprehensive relevance value calculation method, sort the corresponding business items based on the comprehensive relevance value, and push them according to a preset push method;

[0152] The adjustment module 407 is used to record the operation sequence of the user on the pushed business items to obtain user behavior information; obtain the first association value and the second association value and the association value weight of the business item corresponding to the user behavior information to obtain the association value information; calculate the correlation between the user behavior information and the association value information through a preset neural network model to obtain the correlation; adjust the association value weight according to the correlation;

[0153] The second push module 408 is used to extract the first association value and the second association value corresponding to all the business items for each of the users, and sort them to obtain the first push sequence and the second push sequence corresponding to each of the users; select a corresponding number of business items from the first push sequence and the second push sequence according to the preset first quantity threshold and the second quantity threshold; and push the corresponding number of business items to the corresponding user according to the preset push method.

[0154] In this embodiment, the first generating module 402 includes:

[0155] The first parsing unit 4021 is used to parse the business data to obtain numerical business data and category business data;

[0156] A normalization unit 4022, configured to normalize the numerical service data to obtain numerical service data features;

[0157] The encoding unit 4023 is used to perform one-hot encoding on the categorical business data to obtain categorical business data features;

[0158] A feature conversion unit 4024 is used to convert the numerical business data features and the categorical business data features into corresponding feature vectors, and project the feature vectors to preset business dimensions to obtain features of each of the business dimensions;

[0159] A feature statistics unit 4025 is used to count the features of each business dimension according to the business dimension to obtain a statistical feature value corresponding to each business dimension;

[0160] The first generating unit 4026 is used to encapsulate the business dimension, the statistical characteristic value corresponding to the business dimension and the corresponding business data according to a preset format to obtain a business item.

[0161] In this embodiment, the first calculation module 403 includes:

[0162] The second parsing unit 4031 is used to parse the user data to obtain user category information corresponding to each user;

[0163] The determining unit 4032 is configured to query the association weight corresponding to each user in a preset association weight table according to the user category information, wherein the association weight includes a dimension weight corresponding to each business dimension;

[0164] The third parsing unit 4033 is used to parse the business items to obtain statistical characteristic values ​​corresponding to each of the business dimensions;

[0165] The first calculation unit 4034 is used to multiply and accumulate the statistical feature value with the corresponding dimension weight according to the association weight corresponding to each user to obtain a first association value.

[0166] In this embodiment, the second generating module 404 includes:

[0167] The fourth parsing unit 4041 is used to parse the user behavior information to obtain the user's operation information on the historical business items, wherein the operation information includes the associated items corresponding to the operation behavior and the degree of association between the operation behavior and the associated items;

[0168] The second generating unit 4042 is used to store the related items and the correlation degrees according to the corresponding relationship to obtain the user related item set.

[0169] In this embodiment, the first push module 406 includes:

[0170] The second calculation unit 4061 is used to multiply the first association value and the second association value by the corresponding association value weights according to a preset association value weight, and then add them up to obtain a comprehensive association value;

[0171] The sorting unit 4062 is used to extract the comprehensive correlation values ​​corresponding to all business items for each user, and sort them to obtain the item push sequence corresponding to each user;

[0172] The push unit 4063 is used to select a corresponding number of business items from the item push sequence according to a preset push quantity threshold, and push them to corresponding users.

[0173] Through the implementation of the above-mentioned device, by acquiring business data and user data, and extracting features from the business data, business data features are obtained, and business items are generated; user category information corresponding to the user is extracted from the user data, and on this basis, a first association value between the business item and the user is calculated; user behavior information is extracted from the user data, and a set of user-related items is generated; a collaborative filtering recommendation algorithm is called to calculate similarity to obtain a second association value; a comprehensive association value is calculated based on the first association value and the second association value, and the corresponding business items are sorted based on the comprehensive association value, and pushed according to a preset push method; thereby solving the problem of low accuracy of information push in the business information push scenario in the prior art.

[0174] See also Figure 6 , an embodiment of the computer device in the embodiment of the present invention is described in detail below from the perspective of hardware processing.

[0175] Figure 6 6 is a schematic diagram of the structure of a computer device provided by an embodiment of the present invention. The computer device 600 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 610 (for example, one or more processors) and a memory 620, and one or more storage media 630 (for example, one or more mass storage devices) storing application programs 633 or data 632. Among them, the memory 620 and the storage medium 630 can be short-term storage or permanent storage. The program stored in the storage medium 630 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the computer device 600. Furthermore, the processor 610 can be configured to communicate with the storage medium 630 to execute a series of instruction operations in the storage medium 630 on the computer device 600.

[0176] The computer device 600 may also include one or more power supplies 640, one or more wired or wireless network interfaces 650, one or more input and output interfaces 660, and / or one or more operating systems 631, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 6 The computer device structure shown does not constitute a limitation on the computer device provided in the present application, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0177] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the above-mentioned information push method.

[0178] In practical applications, the above-mentioned methods can be implemented based on artificial intelligence technology, where artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. It can be executed based on a server, which can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0179] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0180] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0181] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An information push method, characterized in that: The information pushing method comprises: Obtain business data and user data; Extracting features from the business data to obtain business data features, and generating business items based on the business data features; Parsing the user data to obtain user category information corresponding to the user, determining an association weight corresponding to the user according to the user category information, and calculating a first association value between the business item and the user according to the association weight; Extracting user behavior information corresponding to the user from the user data, matching related items according to the user behavior information, and generating a user related item set corresponding to the user based on the related items; Invoke a collaborative filtering recommendation algorithm to calculate the similarity between the business item and each element in the user-related item set to obtain a second association value; According to a preset comprehensive relevance value calculation method, a comprehensive relevance value is calculated based on the first relevance value and the second relevance value, and corresponding business items are sorted based on the comprehensive relevance value, and pushed according to a preset push method; The extracting features of the business data to obtain business data features, and generating business items based on the business data features, includes: Parsing the business data to obtain numerical business data and category business data; Normalizing the numerical business data to obtain numerical business data features; Performing one-hot encoding on the categorical business data to obtain categorical business data features; Converting the numerical business data features and the categorical business data features into corresponding feature vectors, and projecting the feature vectors to preset business dimensions to obtain features of each of the business dimensions; According to the business dimensions, statistics are collected on the characteristics of each business dimension to obtain statistical characteristic values ​​corresponding to each business dimension; Encapsulate the business dimension, the statistical characteristic value corresponding to the business dimension, and the corresponding business data according to a preset format to obtain a business item; The parsing of the user data to obtain user category information corresponding to the user, determining an association weight corresponding to the user according to the user category information, and calculating a first association value between the business item and the user according to the association weight, includes: Parsing the user data to obtain user category information corresponding to each of the users; According to the user category information, query the association weight corresponding to each user in a preset association weight table, wherein the association weight includes the dimension weight corresponding to each business dimension; Analyze the business items to obtain statistical characteristic values ​​corresponding to each business dimension; According to the association weights corresponding to the users, the statistical feature values ​​are multiplied by the corresponding dimension weights and then accumulated to obtain a first association value; The matching of related items according to the user behavior information and generating a user related item set corresponding to the user based on the related items include: Parsing the user behavior information to obtain the user's operation information on historical business items, wherein the operation information includes the associated items corresponding to the operation behavior and the degree of association between the operation behavior and the associated items; The associated items and the association degrees are stored according to a corresponding relationship to obtain the user associated item set.

2. The information push method according to claim 1, characterized in that: The calculating of the comprehensive relevance value based on the first relevance value and the second relevance value according to the preset comprehensive relevance value calculation method, sorting the corresponding business items based on the comprehensive relevance value, and pushing according to the preset pushing method includes: According to a preset association value weight, multiplying the first association value and the second association value by the corresponding association value weight and adding them together to obtain a comprehensive association value; For each of the users, the comprehensive correlation values ​​corresponding to all business items are extracted and sorted to obtain the item push sequence corresponding to each of the users; According to the preset push quantity threshold, a corresponding number of business items are selected from the item push sequence and pushed to the corresponding users.

3. The information push method according to claim 2, characterized in that: After calculating a comprehensive relevance value based on the first relevance value and the second relevance value according to a preset comprehensive relevance value calculation method, sorting corresponding business items based on the comprehensive relevance value, and pushing them according to a preset push method, the method further includes: Record the user's operation sequence on the pushed business items to obtain user behavior information; Obtain a first correlation value, a second correlation value, and a correlation value weight of a business item corresponding to the user behavior information to obtain correlation value information; Calculating the correlation between the user behavior information and the associated value information through a preset neural network model to obtain a correlation degree; The association value weight is adjusted according to the relevance.

4. The information push method according to any one of claims 1 to 3, characterized in that: After invoking the collaborative filtering recommendation algorithm to calculate the similarity between the business item and each element in the user-related item set to obtain the second association value, the method further includes: For each of the users, first association values ​​and second association values ​​corresponding to all business items are extracted respectively, and sorted to obtain a first push sequence and a second push sequence corresponding to each of the users; According to a preset first quantity threshold and a second quantity threshold, respectively selecting a corresponding number of business items from the first push sequence and the second push sequence; According to the preset push method, the corresponding number of business items are pushed to the corresponding users.

5. An information push device, characterized in that: The device comprises: Acquisition module, used to acquire business data and user data; A first generating module is used to extract features from the business data to obtain business data features, and generate business items based on the business data features; a first calculation module, configured to parse the user data to obtain user category information corresponding to the user, determine an association weight corresponding to the user according to the user category information, and calculate a first association value between the business item and the user according to the association weight; A second generating module is used to extract user behavior information corresponding to the user from the user data, match related items according to the user behavior information, and generate a user related item set corresponding to the user based on the related items; A second calculation module is used to call a collaborative filtering recommendation algorithm to calculate the similarity between the business items and each element in the user-related items set to obtain a second association value; A first push module, configured to calculate a comprehensive relevance value based on the first relevance value and the second relevance value according to a preset comprehensive relevance value calculation method, sort the corresponding business items based on the comprehensive relevance value, and push them according to a preset push method; The first generation module comprises: A first parsing unit, used to parse the business data to obtain numerical business data and category business data; A normalization unit, used to normalize the numerical business data to obtain numerical business data features; An encoding unit, used for performing one-hot encoding on the categorical business data to obtain categorical business data features; A feature conversion unit, used to convert the numerical business data features and the categorical business data features into corresponding feature vectors, and project the feature vectors to preset business dimensions to obtain features of each of the business dimensions; A feature statistics unit, used to count the features of each business dimension according to the business dimension, and obtain a statistical feature value corresponding to each business dimension; A first generating unit is used to encapsulate the business dimension, the statistical characteristic value corresponding to the business dimension and the corresponding business data according to a preset format to obtain a business item; The first calculation module includes: A second parsing unit, used to parse the user data to obtain user category information corresponding to each of the users; A determination unit, configured to query, according to the user category information, the association weight corresponding to each user in a preset association weight table, wherein the association weight includes a dimension weight corresponding to each business dimension; A third parsing unit is used to parse the business items to obtain statistical characteristic values ​​corresponding to each of the business dimensions; A first calculation unit, configured to multiply and accumulate the statistical feature value by the corresponding dimension weight according to the association weight corresponding to each user, to obtain a first association value; The second generation module includes: a fourth parsing unit, used to parse the user behavior information to obtain the user's operation information on historical business items, wherein the operation information includes related items corresponding to the operation behavior and the degree of correlation between the operation behavior and the related items; a second generation unit, used to store the related items and the degree of correlation according to the corresponding relationship to obtain the user related item set.

6. A computer device, characterized in that: include: A memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected via a line; The at least one processor calls the instructions in the memory to enable the computer device to execute each step of the information push method as described in any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, each step of the information push method according to any one of claims 1 to 4 is implemented.

Images