Knowledge recommendation method and device, computer equipment and readable storage medium

Abstract

The invention relates to a knowledge recommendation method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring a knowledge attribute data set of each knowledge, a target historical behavior data set of a target user and a target user attribute data set; screening each knowledge based on the target historical behavior data set to obtain each initial target knowledge, and constructing each knowledge user attribute data set according to the knowledge attribute data set of each initial target knowledge and the target user attribute data set; and according to the knowledge recommendation model and the knowledge user attribute data sets, determining the integrating degree of each piece of initial target knowledge and the target user, and according to each integrating degree, determining target recommendation knowledge in each piece of initial target knowledge. By adopting the method, the accuracy of the knowledge recommendation method can be improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a knowledge recommendation method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Art

[0002] In the field of R&D, the acquisition and sharing of knowledge is the key to promoting innovation and improving work efficiency. Each R&D personnel (user) will browse the knowledge on the R&D platform in their spare time to acquire new R&D knowledge. The R&D platform will recommend target knowledge to target users based on the knowledge recommendation method.

[0003] The current knowledge recommendation method obtains the interactive data set of each knowledge. Each interactive data set contains the number of views, likes and comments of the knowledge. Each knowledge is sorted according to the number of comments, likes and views, and a comment knowledge list, a like knowledge list and a browse knowledge list are obtained. The target recommended knowledge is determined in the comment knowledge list, the like knowledge list and the browse knowledge list according to the preset display quantity.

[0004] However, current knowledge recommendation methods only recommend target knowledge to users based on interactive datasets of knowledge, with a single factor considered, which leads to low accuracy of target knowledge recommendation. Summary of the invention

[0005] Based on this, it is necessary to provide a knowledge recommendation method, apparatus, computer device, computer-readable storage medium and computer program product to address the above technical issues.

[0006] In a first aspect, the present application provides a knowledge recommendation method, comprising:

[0007] Obtaining knowledge attribute data sets of each knowledge, target historical behavior data sets of target users, and target user attribute data sets;

[0008] Filtering each of the knowledge based on the target historical behavior data set to obtain each initial target knowledge, and constructing each knowledge user attribute data set based on the knowledge attribute data set of each initial target knowledge and the target user attribute data set;

[0009] The degree of fit between each of the initial target knowledge and the target user is determined according to the knowledge recommendation model and the attribute data sets of each knowledge user, and the target recommended knowledge is determined in each of the initial target knowledge according to the degree of fit.

[0010] In one embodiment, the screening of each of the knowledge based on the target historical behavior data set to obtain each of the initial target knowledge includes:

[0011] For each piece of knowledge, determining whether a knowledge identifier of the knowledge exists in the target historical behavior data set;

[0012] If the knowledge identifier of the knowledge does not exist in the target historical behavior data set, the knowledge is determined as the initial target knowledge.

[0013] In one embodiment, before determining the degree of fit between each of the initial target knowledge and the target user based on the knowledge recommendation model and each knowledge user attribute data set, the method further includes:

[0014] Acquire a user attribute data set and a historical behavior data set of each user, and construct each initial sample data set according to each of the knowledge attribute data sets, each of the user attribute data sets, and each of the historical behavior data sets;

[0015] Update each of the initial sample data sets according to the fit scoring rule to obtain each sample data set;

[0016] A natural language processing model is trained based on each of the sample data sets until the natural language processing model meets a preset training stop condition, and the natural language processing model is determined as a knowledge recommendation model.

[0017] In one embodiment, the constructing each initial sample data set according to each knowledge attribute data set, each user attribute data set and each historical behavior data set includes:

[0018] For each of the historical behavior data sets of the user, determining a knowledge attribute data set corresponding to each historical behavior data subset in the historical behavior data set;

[0019] An initial sample data set is constructed according to the user attribute data set, the historical behavior data subset, the user behavior type and the knowledge attribute data set.

[0020] In one embodiment, the updating of each of the initial sample data sets according to the fit scoring rule to obtain each sample data set includes:

[0021] Determine a fit score corresponding to each of the initial sample data sets according to a fit score rule;

[0022] The fit score is added to the initial sample data set to obtain a sample data set.

[0023] In one embodiment, determining target recommendation knowledge from each of the initial target knowledge according to each of the fit degrees includes:

[0024] Sorting the initial target knowledge in descending order of their fit, to obtain an initial target knowledge sequence;

[0025] The target recommended knowledge is determined in the initial target knowledge sequence according to a preset display quantity.

[0026] In a second aspect, the present application also provides a knowledge recommendation device, comprising:

[0027] An acquisition module is used to acquire a knowledge attribute data set of each knowledge, a target historical behavior data set of a target user, and a target user attribute data set;

[0028] A screening module, used for screening each of the knowledge based on the target historical behavior data set to obtain each initial target knowledge, and constructing each knowledge user attribute data set according to the knowledge attribute data set of each initial target knowledge and the target user attribute data set;

[0029] The determination module is used to determine the degree of fit between each of the initial target knowledge and the target user according to the knowledge recommendation model and each knowledge user attribute data set, and determine the target recommended knowledge in each of the initial target knowledge according to each of the degrees of fit.

[0030] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0031] Obtaining knowledge attribute data sets of each knowledge, target historical behavior data sets of target users, and target user attribute data sets;

[0032] Filtering each of the knowledge based on the target historical behavior data set to obtain each initial target knowledge, and constructing each knowledge user attribute data set based on the knowledge attribute data set of each initial target knowledge and the target user attribute data set;

[0033] The degree of fit between each of the initial target knowledge and the target user is determined according to the knowledge recommendation model and the attribute data sets of each knowledge user, and the target recommended knowledge is determined in each of the initial target knowledge according to the degree of fit.

[0034] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0035] Obtaining knowledge attribute data sets of each knowledge, target historical behavior data sets of target users, and target user attribute data sets;

[0036] Filtering each of the knowledge based on the target historical behavior data set to obtain each initial target knowledge, and constructing each knowledge user attribute data set based on the knowledge attribute data set of each initial target knowledge and the target user attribute data set;

[0037] The degree of fit between each of the initial target knowledge and the target user is determined according to the knowledge recommendation model and the attribute data sets of each knowledge user, and the target recommended knowledge is determined in each of the initial target knowledge according to the degree of fit.

[0038] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0039] Obtaining knowledge attribute data sets of each knowledge, target historical behavior data sets of target users, and target user attribute data sets;

[0040] Filtering each of the knowledge based on the target historical behavior data set to obtain each initial target knowledge, and constructing each knowledge user attribute data set based on the knowledge attribute data set of each initial target knowledge and the target user attribute data set;

[0041] The degree of fit between each of the initial target knowledge and the target user is determined according to the knowledge recommendation model and the attribute data sets of each knowledge user, and the target recommended knowledge is determined in each of the initial target knowledge according to the degree of fit.

[0042] The above-mentioned knowledge recommendation method, device, computer equipment, computer-readable storage medium and computer program product obtain the knowledge attribute data set of each knowledge, the target historical behavior data set and the target user attribute data set of the target user; screen each of the knowledge based on the target historical behavior data set to obtain each initial target knowledge, and construct each knowledge user attribute data set based on the knowledge attribute data set and the target user attribute data set of each initial target knowledge; determine the degree of fit between each of the initial target knowledge and the target user based on the knowledge recommendation model and each knowledge user attribute data set, and determine the target recommended knowledge in each of the initial target knowledge based on each degree of fit. According to this method, each knowledge is screened through the historical behavior data set to obtain each initial target knowledge that the target user has not browsed. Then, the degree of fit of the initial target knowledge for the target user is determined through the knowledge attribute data set, the target user attribute data set and the knowledge recommendation model of each initial target knowledge, and the target recommended knowledge is determined based on each degree of fit, and the target user attribute data set, the knowledge recommendation model and the target knowledge attribute data are added. The personalized recommendation of the target recommended knowledge of the target user is realized, the accuracy of the target recommended knowledge is improved, and the accuracy of the knowledge recommendation method is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 is a flowchart of a knowledge recommendation method in one embodiment;

[0045] Figure 2 A schematic diagram of a process for determining initial target knowledge in one embodiment;

[0046] Figure 3 A schematic diagram of a process for determining a knowledge recommendation model in one embodiment;

[0047] Figure 4 A schematic diagram of a process for constructing an initial sample data set in one embodiment;

[0048] Figure 5 A schematic diagram of a process for determining each slave sample data set in one embodiment;

[0049] Figure 6 A schematic diagram of a process for determining target recommendation knowledge in one embodiment;

[0050] Figure 7 A schematic diagram of a flow chart of a knowledge recommendation method in another embodiment;

[0051] Figure 8 is a structural block diagram of a knowledge recommendation device in one embodiment;

[0052] Fig. 9 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0054] In one embodiment, Figure 1 As shown, a knowledge recommendation method is provided. The embodiment of the present application takes the method applied to a computer device as an example for explanation. The embodiment of the present application does not limit the execution device for executing the knowledge recommendation method, and includes the following steps:

[0055] Step 102: Acquire a knowledge attribute data set of each piece of knowledge, a target historical behavior data set of a target user, and a target user attribute data set.

[0056] In implementation, the computer device obtains the knowledge attribute data set of each knowledge from the database. The knowledge attribute data set includes the basic attribute data subset of the knowledge, the interactive data subset of the knowledge, and the author attribute data subset of the knowledge. At the same time, the computer device obtains the target historical behavior data set and the user attribute data set of the target user from the database. The target user is the user of the knowledge to be recommended. The target historical behavior data set is the historical record data set of the target user's historical browsing, commenting or liking of the knowledge. The target user attribute data of the target user includes the basic attribute data of the target user.

[0057] Specifically, when the target user wants to acquire knowledge, he will log in to the knowledge platform. After the target user logs in to the knowledge platform, the computer device obtains the target user ID of the target user, and queries and obtains the target historical behavior data set and the target user attribute data set of the target user in the database according to the target user ID. At the same time, the computer device obtains the knowledge attribute data set of the full knowledge from the database. The knowledge attribute data set includes a basic attribute data subset, an interactive data subset, and an author attribute data subset. The basic attribute data subset represents the basic attributes of the knowledge; the interactive data subset represents the interaction between each user and the knowledge; the author attribute data subset represents the basic attributes of the author who created the knowledge. The basic attribute data subset of the knowledge includes the specific content, release time, length of the knowledge, and whether the knowledge contains a cover, etc. The interactive data subset of the knowledge includes the number of views, likes, comments, and comments of the knowledge, and the author attribute data subset of the knowledge includes the author ID, author name, author historical creation knowledge, author creation level, etc. of the author who created the knowledge. The target historical behavior data set includes historical behavior data subsets of three behaviors of the user, namely, the historical browsing data subset, the historical likes data subset, and the historical comment data subset. The historical browsing data subset contains the knowledge records of each browsing history of the target user. The historical likes data subset contains the knowledge records of each likes history of the target user. The historical comments data subset contains the knowledge records of each comment history of the target user. The target user attribute data set contains the target user ID, target user name, user tags, historical creation records, etc. of the target user.

[0058] Step 104 , screening each piece of knowledge based on the target historical behavior data set to obtain each initial target knowledge, and constructing each knowledge user attribute data set based on the knowledge attribute data set of each initial target knowledge and the target user attribute data set.

[0059] In implementation, the computer device selects each initial target knowledge that has no interaction with the target user from each knowledge according to the target historical behavior data set of the target user. Then, the computer device constructs a knowledge user attribute data set according to the knowledge attribute data set and the target user attribute data set of each initial target knowledge.

[0060] Specifically, the target historical behavior data set includes a subset of historical browsing data. The computer device screens each initial target knowledge that has no interaction with the target user from each knowledge based on the subset of historical browsing data of the target user. The initial target knowledge is the knowledge to be recommended to the target user. Then, the computer device combines the knowledge attribute data set and the target user attribute data set of each initial target knowledge to obtain a knowledge user attribute data set. The knowledge attribute data includes the attribute characteristics of the initial target knowledge. The target user attribute data set includes the attribute characteristics of the target user. By combining the knowledge attribute data set and the target user attribute data set, the knowledge user attribute data set includes the attribute characteristics of the target user and the attribute characteristics of the knowledge.

[0061] Step 106, determining the degree of fit between each initial target knowledge and the target user according to the knowledge recommendation model and the attribute data set of each knowledge user, and determining target recommended knowledge from each initial target knowledge according to each degree of fit.

[0062] Among them, the knowledge recommendation model is obtained by training a natural language processing model based on a sample data set containing a user attribute data set, a historical behavior data subset, a user behavior type, a knowledge attribute data set, and a fit score.

[0063] In implementation, the computer device inputs the knowledge user attribute data set of each initial target knowledge into the knowledge recommendation model, and processes the knowledge user attribute data set through the knowledge recommendation model to obtain the degree of fit between the target user and the initial target knowledge. Then, the computer device determines each target recommended knowledge in each initial target knowledge according to the degree of fit of each initial target knowledge. The computer device constructs a knowledge recommendation webpage based on each target recommended knowledge, and displays the knowledge recommendation webpage, thereby recommending the target recommended knowledge to the target user.

[0064] Specifically, the computer device constructs each sample data set based on each user attribute data set, each historical behavior data subset, each historical behavior type, each knowledge attribute data set and each fit score. The user attribute data set represents the user attribute characteristics. The historical behavior data subset represents the user's behavior characteristics, and the behavior characteristics represent the user's interest. The knowledge attribute data set represents the attributes of the knowledge. Training the natural processing model with each sample data set enables the natural language processing model to learn the association between the user's attribute characteristics, behavior characteristics, interest characteristics and the attribute characteristics of the knowledge, and allows the natural language processing model to learn the behavior characteristics and interest characteristics of each user. The association relationship is the fit between the knowledge and the user. Then, the computer device inputs the knowledge user attribute data set containing the user attribute characteristics of the target user and the knowledge attribute characteristics of the initial target knowledge into the knowledge recommendation model, and determines the association between the user attribute characteristics and the knowledge attribute model through the knowledge recommendation model, thereby obtaining the fit between the target user and the initial target knowledge. The computer device determines the target recommended knowledge of the target user in each initial target knowledge according to the fit of each initial target knowledge.

[0065] In the above-mentioned knowledge recommendation method, each knowledge is screened through the historical behavior data set to obtain each initial target knowledge that the target user has not browsed. Then, the initial target knowledge is determined to be suitable for the target user through the knowledge attribute data set, the target user attribute data set and the knowledge recommendation model of each initial target knowledge, and the target recommended knowledge is determined based on each degree of fit. The target user attribute data set, the knowledge recommendation model and the target knowledge attribute data are added to the influencing factors, and the personalized recommendation of the target recommended knowledge for the target user is realized, the accuracy of the target recommended knowledge is improved, and the accuracy of the knowledge recommendation method is further improved.

[0066] In an exemplary embodiment, Figure 2 As shown, the specific process of filtering each knowledge based on the target historical behavior data set in step 104 to obtain each initial target knowledge includes steps 202 to 204. Among them:

[0067] Step 202: for each piece of knowledge, determine whether the knowledge identifier of the knowledge exists in the target historical behavior data set.

[0068] Among them, the target historical behavior data set contains historical behavior data subsets of three behaviors of the target user, namely, the historical browsing data subset, the historical like data subset and the historical comment data subset.

[0069] In implementation, the computer device determines whether the knowledge identifier of each knowledge exists in the target historical behavior data set. If the knowledge identifier of the knowledge exists in the target historical behavior data set, the computer device determines that the knowledge has interacted with the target user and filters out the knowledge.

[0070] Specifically, the computer device determines, for each piece of knowledge, whether there is a knowledge identifier of the knowledge in each historical browsing record in the historical browsing data subset. Since users need to browse the knowledge before they can interact with the knowledge by liking, commenting, etc., therefore, by determining whether there is a knowledge identifier of the knowledge in the historical browsing data subset, it is possible to determine whether the target user has interacted with the knowledge.

[0071] In an optional embodiment, the computer device determines, for each piece of knowledge, whether there is a knowledge identifier of the knowledge in the historical browsing data subset, the historical like data subset, and the historical comment data subset. If there is a knowledge identifier of the knowledge in the historical browsing data subset, the historical like data subset, or the historical comment data subset, the computer device filters out the knowledge.

[0072] Step 204: If the knowledge identifier of the knowledge does not exist in the target historical behavior data set, the knowledge is determined as the initial target knowledge.

[0073] In implementation, if the knowledge identifier of the knowledge does not exist in the target historical behavior data set, the computer device determines that the knowledge does not interact with the target user and determines the knowledge as the initial target knowledge. That is, the computer device determines the knowledge as the knowledge to be recommended to the target user.

[0074] In this embodiment, each knowledge is filtered through the target historical behavior data set, and each initial target knowledge is filtered to obtain each initial target knowledge that has no interaction with the target user, which facilitates the subsequent determination of target recommendation knowledge in each initial target knowledge, realizes personalized recommendation for users, and improves the accuracy of the knowledge recommendation method.

[0075] In an exemplary embodiment, before determining the degree of fit based on the knowledge recommendation model, it is necessary to first train a natural language processing model based on each sample data set to obtain a knowledge recommendation model. Figure 3 As shown, before step 206, the specific processing process of the knowledge recommendation method also includes steps 302 to 306. Among them:

[0076] Step 302: Acquire the user attribute data set and the historical behavior data set of each user, and construct each initial sample data set according to each knowledge attribute data set, each user attribute data set and each historical behavior data set.

[0077] In implementation, the computer device obtains user attribute data sets and historical behavior data sets of all users from the database, and then constructs each initial sample data set based on each knowledge attribute data set, each user attribute data set and each historical behavior data set.

[0078] Specifically, the computer device obtains the user data set and historical behavior data set of all users from the database. The historical behavior data set of each user contains various historical behavior data subsets. Then, the computer device determines the knowledge attribute data set corresponding to the historical behavior data subset for each user's historical behavior data subset, and constructs an initial sample data set based on the user attribute data set, the historical behavior data subset, and the knowledge attribute data corresponding to the historical behavior data subset.

[0079] Step 304: Update each initial sample data set according to the fit scoring rule to obtain each sample data set.

[0080] In implementation, the computer device determines the fit score of each initial sample data set according to the fit score rule, and adds the fit score to the initial sample data set to obtain the sample data set.

[0081] Specifically, the initial sample data set includes a user attribute data subset, a historical behavior data subset, a user behavior type, and knowledge attribute data corresponding to the historical behavior data subset. The computer device determines the degree of association between the user and the knowledge in the initial sample data set according to the fit score rule to obtain a fit score. Then, the computer device adds the fit score to the initial sample data set to obtain a sample data set.

[0082] Step 306: train the natural language processing model based on each sample data set until the natural language processing model meets a preset training stop condition, and determine the natural language processing model as a knowledge recommendation model.

[0083] In the implementation, the computer device determines the predicted fit between the user and the knowledge in the sample data set based on the sample data set and the natural language model for each sample data set. Then, the computer device determines whether the natural language processing model meets the preset training stop condition based on each predicted fit and each fit score. When the natural language processing model meets the training stop condition, the computer device determines the natural language processing model as a knowledge recommendation model. The natural language processing model is trained by the sample data set so that the natural language processing model can learn the behavioral characteristics of each user, and learn the correlation between the attribute characteristics of the user, the behavioral characteristics and the attribute characteristics of the knowledge.

[0084] Specifically, the computer device inputs the sample data set into the natural language processing model for each sample data set, and processes the sample data set through the natural language processing model to obtain the predicted fit between the user and the knowledge in the sample data set. Then, the computer device performs data operations on each predicted fit and each fit according to the loss function algorithm to obtain a loss value. The computer device determines whether the loss value reaches a preset loss threshold. If the loss value reaches the preset loss threshold, the computer device determines that the natural language processing model meets the preset training stop condition. The computer device determines the natural language processing model as a knowledge recommendation model. If the loss value does not reach the loss threshold, the computer device determines that the natural language processing model does not meet the preset training stop condition, and updates the parameters of the natural language processing model according to the loss value, so that the natural language processing model can learn the behavior characteristics of each user, and learn the association between the attribute characteristics of the user, the behavior characteristics and the attribute characteristics of the knowledge. Then, the computer device continues to execute the step of inputting the sample data set into the natural language processing model for each sample data set until the natural language processing model meets the training stop condition.

[0085] Optionally, the training stop condition may be, but is not limited to, the loss value reaching a preset loss threshold, or the number of training rounds reaching a preset training round threshold. The loss threshold and the training round threshold are determined according to the training requirements, and the embodiment of the present application does not limit the training conditions.

[0086] In this embodiment, by constructing various sample data sets, a sample data set including user attribute characteristics, behavioral characteristics and knowledge attribute characteristics is obtained, and a natural language processing model is trained based on the sample data set so that a language recommendation model is enabled, so that the language model can learn the behavioral characteristics of each user, and learn the relationship between behavioral characteristics, user attribute characteristics and knowledge attribute characteristics, so as to facilitate the subsequent recommendation of knowledge to users based on the learned content, thereby improving the accuracy of the knowledge recommendation model and thereby improving the accuracy of the knowledge recommendation method.

[0087] In an exemplary embodiment, Figure 4 As shown, the specific processing process of constructing each initial sample data set according to each knowledge attribute data set, each user attribute data set and each historical behavior data set in step 302 includes steps 402 to 404. Among them:

[0088] Step 402: for each user's historical behavior data set, determine the knowledge attribute data set corresponding to each historical behavior data subset in the historical behavior data set.

[0089] The historical behavior data set includes various historical behavior data subsets, and each historical behavior data subset includes the identifier of the behavior object, that is, the knowledge identifier.

[0090] In implementation, the computer device determines the knowledge attribute data set corresponding to each historical behavior data subset in each knowledge attribute data set according to the knowledge identifier in the historical behavior data subset for each historical behavior data subset in each user's historical behavior data set. By determining the knowledge attribute data set corresponding to each historical behavior data subset, the knowledge of historical interaction of the target user is determined.

[0091] Specifically, the historical behavior data subset is divided into a historical browsing data subset, a historical like data subset, and a historical comment data subset. For each historical browsing data subset in the historical behavior data set of each user, the computer device determines the knowledge attribute data set corresponding to the historical browsing data subset in each knowledge attribute data set according to the knowledge identifier of the knowledge of the user's historical browsing in the historical browsing data subset. The knowledge attribute data set corresponding to the historical browsing data subset is the knowledge attribute data set of the knowledge of the user's historical browsing. Then, for each historical like data subset in the historical behavior data set of each user, the computer device determines the knowledge attribute data set corresponding to the historical like data subset in each knowledge attribute data set according to the knowledge identifier of the knowledge of the user's historical likes in the historical like data subset. The knowledge attribute data set corresponding to the historical like data subset is the knowledge attribute data set of the knowledge of the user's historical likes. For each historical comment data subset in the historical behavior data set of each user, the computer device determines the knowledge attribute data set corresponding to the historical comment data subset in each knowledge attribute data set according to the knowledge identifier of the knowledge of the user's historical comments in the historical comment data subset. The knowledge attribute data set corresponding to the historical comment data subset is the knowledge attribute data set of the knowledge of the user's historical comments.

[0092] Step 404: construct an initial sample data set based on the user attribute data set, the historical behavior data subset, the user behavior type and the knowledge attribute data set.

[0093] In implementation, the computer device constructs an initial sample data set for each historical behavior data subset based on the historical behavior data subset, the knowledge attribute data set corresponding to the historical behavior data subset, the user behavior type represented by the historical behavior data subset, and the knowledge attribute data set.

[0094] Specifically, if the historical behavior data subset is a historical browsing data subset, then the user behavior represented by the historical browsing data subset is browsing behavior. The computer device combines the historical browsing data subset, the knowledge attribute data corresponding to the historical browsing data subset, the browsing behavior, and the user attribute data set to obtain an initial sample data set. If the historical behavior data subset is a historical likes data subset, then the user behavior represented by the historical likes data subset is likes behavior. The computer device combines the historical likes data subset, the knowledge attribute data corresponding to the historical likes data subset, the likes behavior, and the user attribute data set to obtain an initial sample data set. If the historical behavior data subset is a historical comment data subset, then the user behavior represented by the historical comment data subset is comment behavior. The computer device combines the historical comment data subset, the knowledge attribute data corresponding to the historical comment data subset, the comment behavior, and the user attribute data set to obtain an initial sample data set.

[0095] In this embodiment, by determining the knowledge attribute data set corresponding to each historical behavior data subset, the knowledge attribute features of the user interaction are obtained. Then, based on the user attribute data set, the historical behavior data subset, the user behavior type and the knowledge attribute data set, an initial sample data set is constructed, and an initial sample data set containing user attribute features, knowledge attribute features and user behavior features is obtained, which is convenient for subsequent training of the natural language processing model.

[0096] In an exemplary embodiment, Figure 5 As shown, the specific processing process of step 304 includes steps 502 to 504. Among them:

[0097] Step 502: Determine the fit score corresponding to each initial sample data set according to the fit score rule.

[0098] In implementation, the computer device is pre-set with a fit scoring rule, and the computer device processes each initial sample data set according to the fit scoring rule to obtain a fit score corresponding to the initial sample data set.

[0099] Specifically, the fit scoring rules include knowledge scoring rules and interaction scoring rules. The initial sample data set includes a user attribute data set, a historical behavior data subset, a user behavior type, and knowledge attribute data. The computer device determines the basic score of the knowledge according to the knowledge scoring rules and the knowledge attribute data set, and determines the interaction score between the user and the knowledge according to the interaction scoring rules, the historical behavior data subset, the user behavior type, and the user attribute data set. The basic score represents the quality of the knowledge. The higher the basic score, the higher the quality of the knowledge, and the lower the basic score, the lower the quality of the knowledge. The interaction score represents the user's interest in the knowledge. The higher the interaction score, the more interested the user is in the knowledge; the lower the interaction score, the less interested the user is in the knowledge. Then, the computer device performs weighted processing on the interaction score and the basic score according to the preset weights to obtain the fit score between the user and the knowledge in the initial sample data set.

[0100] In an exemplary embodiment, the knowledge scoring rules include each like number range and the like score corresponding to each like number range, each comment number range and the comment score corresponding to each comment number range, each browse number range and the browse score corresponding to each browse number range, comment sentiment score rules, each knowledge word number range and the word number score corresponding to each knowledge word number range, each author creation level and the author score and cover score corresponding to each author creation level. The knowledge attribute data set includes the number of likes, comment number, browse number, comment content, author level, knowledge word number and cover score of the knowledge.

[0101] The computer device determines the target like count range where the number of likes falls, and determines the target like score corresponding to the target like count range. The computer device determines the target comment count range where the number of comments falls, and determines the target comment score corresponding to the target comment count range. The computer device determines the target browse count range where the number of views falls, and determines the target browse score corresponding to the target browse count range. The computer device determines the target knowledge word count range where the number of knowledge words falls, and determines the target word count score corresponding to the target knowledge word count range. The computer device determines the sentiment score based on the comment sentiment score rule and the comment content, and determines the cover score based on whether there is a cover in the knowledge attribute data set. The computer device performs weighted processing on the target like score, target comment score, target browse score, target word count score, sentiment score and cover score according to the preset knowledge score weight to obtain a basic score.

[0102] The interactive scoring rule includes each browsing time range and the browsing time score, collection action score, author attention score and comment action score corresponding to each browsing time range. The historical behavior data subset and user behavior type include the time when the user browses the knowledge, the collection action field, the comment action field and the author identification of the knowledge. The user attribute data set includes a list of author identifications that the user pays attention to. The computer device determines the time when the user browses the knowledge to determine the target browsing time range, and determines the target browsing time score corresponding to the browsing time range. Then, the computer device determines the user collection score according to the collection action field and the collection action score, and determines the user comment score according to the comment action score and the comment action field. The computer device determines whether the user pays attention to the author of the knowledge according to the author identification list that the user pays attention to and the author identification of the knowledge, and obtains the attention result. Then, the computer device determines the user's attention score according to the attention result and the author attention score. The computer device performs weighted processing on the target browsing time score, the user collection score, the user comment score and the user's attention score according to the interactive weight to obtain the interactive score. Then, the computer device performs weighted processing on the interactive score and the basic score according to the preset weight to obtain the fit score of the user and the knowledge in the initial sample data set.

[0103] In an optional embodiment, the computer device determines the number of likes, the number of comments, the content of comments, the browsing time of the user browsing the knowledge, the collection field and the attention result of the knowledge in the initial sample data set. The fit scoring rule includes the comment sentiment score rule, each like number range and the like score corresponding to each like number range, each comment number range and the comment score corresponding to each comment number range, the browsing time score corresponding to each browsing time range, the collection action score and the author attention score. The computer device determines the target like number range where the like number is located, and determines the target like score corresponding to the target like number range. The computer device determines the target comment number range where the comment number is located, and determines the target comment score corresponding to the target comment number range. Then, the computer device determines the sentiment score according to the comment sentiment score rule and the comment content. The computer device determines the comment score according to the target comment score and the sentiment score. Then, the computer device determines the time when the user browses the knowledge to determine the target browsing time range, and determines the target browsing time score corresponding to the browsing time range, and determines the user's author attention score according to the attention result and the author attention score. The computer device determines the user's collection score according to the collection action field and the collection action score. The computer device performs weighted processing on the target like score, comment score, target browsing time score, user collection score and follow-up author score according to the preset weights to obtain the fit score. The weighted processing process is shown in the following formula (1):

[0104] Compatibility score = w1×target likes score + w2×comment score + w3×target browsing time score + w4×user collection score + w5×author follow score (1)

[0105] Among them, in the above formula (1), w1,w2,w3,w4,w5w1,w2,w3,w4,w5 are weight coefficients.

[0106] In particular, it should be noted that the more likes a piece of knowledge has, the higher the target like score will be; the more comments it has, the higher the comment score will be; the more browsing time it has, the higher the target browsing time score will be; users’ collection of knowledge will increase their score, and following the author of the knowledge will also increase their score.

[0107] Step 504: Add the fit score to the initial sample data set to obtain a sample data set.

[0108] In implementation, the computer device adds the fit score to the initial sample data set to obtain the sample data set, thereby clarifying the fit between the knowledge in the sample data set and the user.

[0109] In an exemplary embodiment, the computer device creates a new column and stores each fit score in the column. Then, the computer device adds the fit score to the initial sample data set to obtain the sample data set.

[0110] In this embodiment, the fit score of each initial sample data set is determined by the fit score rule, and the correlation and evaluation criteria between the user attribute features, user behavior features and knowledge attribute features are clarified. Then, the fit score and the initial sample data set are used to construct a sample data set, so that the natural language processing model can learn the correlation between the user attribute features, user behavior features and knowledge attribute features when training the natural language processing model later.

[0111] In an exemplary embodiment, Figure 6 As shown, the specific process of determining the target recommended knowledge from each initial target knowledge according to each degree of fit in step 106 includes steps 602 to 604. Among them:

[0112] Step 602, sorting the initial target knowledge in descending order of their degree of fit to obtain an initial target knowledge sequence.

[0113] In implementation, the computer device sorts each initial target knowledge in order from high to low in terms of the degree of fit of each initial target knowledge to obtain an initial target knowledge sequence.

[0114] Step 604: determine target recommended knowledge in the initial target knowledge sequence according to the preset display quantity.

[0115] In implementation, the display quantity is preset in the computer device. The computer device determines the initial target knowledge of the display quantity in the target knowledge sequence, and determines the initial target knowledge of the display quantity as the target recommended knowledge.

[0116] In an optional embodiment, a compatibility threshold and a display quantity are pre-set in the computer device. According to the compatibility threshold, the computer device selects each target knowledge whose compatibility is greater than the compatibility threshold from each initial target knowledge. Then, the computer device sorts each target knowledge in descending order according to the compatibility of each target knowledge to obtain a target knowledge sequence. The computer device determines the target recommended knowledge in the target knowledge sequence according to the preset display quantity.

[0117] In this embodiment, the target recommendation knowledge is determined in each initial target knowledge through each degree of fit, and the target recommendation knowledge that best fits the target user is obtained, thereby realizing personalized recommendation for the target user, improving the accuracy of the knowledge recommendation method, and enhancing the interactivity and stickiness between the user and the R&D platform.

[0118] In an exemplary embodiment, Figure 7 FIG. 1 is a flow chart of a knowledge recommendation method in an embodiment. Figure 7 As shown, the processing flow of the knowledge recommendation method includes:

[0119] Step 701: Obtain a knowledge attribute data set of all knowledge from the database of the R&D platform, wherein the knowledge attribute data set includes the release time of the knowledge, the length of the knowledge, and whether it contains a cover, etc.

[0120] Step 702: Obtain a historical behavior data set of all users from the database of the R&D platform. The historical behavior data set includes the number of views, likes, comments, and favorites of users.

[0121] Step 703: Based on each user behavior data set and each knowledge attribute data set, a natural language processing model is trained so that the natural language learning model learns user preferences and the degree of association between the user and knowledge, thereby obtaining a knowledge recommendation model.

[0122] Step 704: construct a knowledge user attribute data set based on the target historical behavior data set and the target user attribute data set, and determine the fit between each initial target knowledge and the target user according to the knowledge user attribute data set and the knowledge recommendation model.

[0123] Step 705 , determining target recommended knowledge from each initial target knowledge according to each degree of fit.

[0124] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0125] Based on the same inventive concept, the embodiment of the present application also provides a knowledge recommendation device for implementing the knowledge recommendation method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more knowledge recommendation device embodiments provided below can refer to the limitations on the knowledge recommendation method above, and will not be repeated here.

[0126] In an exemplary embodiment, Figure 8 As shown, a knowledge recommendation device 800 is provided, comprising: an acquisition module 801, a screening module 802 and a determination module 803, wherein:

[0127] The acquisition module 801 is used to acquire the knowledge attribute data set of each knowledge, the target historical behavior data set of the target user and the target user attribute data set.

[0128] The screening module 802 is used to screen each piece of knowledge based on the target historical behavior data set to obtain each initial target knowledge, and to construct each knowledge user attribute data set based on the knowledge attribute data set of each initial target knowledge and the target user attribute data set.

[0129] The determination module 803 is used to determine the degree of fit between each initial target knowledge and the target user according to the knowledge recommendation model and each knowledge user attribute data set, and determine the target recommended knowledge in each initial target knowledge according to each degree of fit.

[0130] In an exemplary embodiment, the screening module 802 includes a first screening submodule and a first building submodule. The first screening submodule includes:

[0131] The first judgment submodule is used to judge, for each piece of knowledge, whether there is a knowledge identifier of the knowledge in the target historical behavior data set.

[0132] The first determination submodule is used to determine the knowledge as the initial target knowledge if the knowledge identifier of the knowledge does not exist in the target historical behavior data set.

[0133] In an exemplary embodiment, the knowledge recommendation device 800 further includes:

[0134] The second acquisition module is used to acquire the user attribute data set and the historical behavior data set of each user, and construct each initial sample data set according to each knowledge attribute data set, each user attribute data set and each historical behavior data set.

[0135] The updating module is used to update each initial sample data set according to the fit scoring rule to obtain each sample data set.

[0136] The training module is used to train the natural language processing model based on each sample data set until the natural language processing model meets the preset training stop condition, and determine the natural language processing model as a knowledge recommendation model.

[0137] In an exemplary embodiment, the second acquisition module includes a first acquisition submodule and a second construction submodule. The second construction submodule includes:

[0138] The second determination submodule is used to determine, for each user's historical behavior data set, a knowledge attribute data set corresponding to each historical behavior data subset in the historical behavior data set.

[0139] The third construction submodule is used to construct an initial sample data set based on the user attribute data set, the historical behavior data subset, the user behavior type and the knowledge attribute data set.

[0140] In an exemplary embodiment, the update module includes:

[0141] The third determination submodule is used to determine the fit score corresponding to each initial sample data set according to the fit score rule.

[0142] The first adding submodule is used to add the fit score to the initial sample data set to obtain a sample data set.

[0143] In an exemplary embodiment, the determination module includes a fourth determination submodule and a fifth determination submodule. The fifth determination submodule includes:

[0144] The sorting submodule is used to sort the initial target knowledge in descending order of their fit, so as to obtain an initial target knowledge sequence.

[0145] The sixth determination submodule is used to determine target recommended knowledge in the initial target knowledge sequence according to a preset display quantity.

[0146] Each module in the above-mentioned knowledge recommendation device can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each module.

[0147] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Fig. 9 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, a knowledge recommendation method is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0148] Those skilled in the art will understand that Fig. 9 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0149] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0150] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0151] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0152] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0153] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0154] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0155] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A knowledge recommendation method, characterized in that: The method comprises: Obtaining knowledge attribute data sets of each knowledge, target historical behavior data sets of target users, and target user attribute data sets; Filtering each of the knowledge based on the target historical behavior data set to obtain each initial target knowledge, and constructing each knowledge user attribute data set based on the knowledge attribute data set of each initial target knowledge and the target user attribute data set; The degree of fit between each of the initial target knowledge and the target user is determined according to the knowledge recommendation model and the attribute data sets of each knowledge user, and the target recommended knowledge is determined in each of the initial target knowledge according to the degree of fit.

2. The method according to claim 1, characterized in that The step of screening each of the knowledge based on the target historical behavior data set to obtain each of the initial target knowledge includes: For each piece of knowledge, determining whether a knowledge identifier of the knowledge exists in the target historical behavior data set; If the knowledge identifier of the knowledge does not exist in the target historical behavior data set, the knowledge is determined as the initial target knowledge.

3. The method according to claim 1, characterized in that Before determining the degree of fit between each of the initial target knowledge and the target user based on the knowledge recommendation model and each knowledge user attribute data set, the method further includes: Acquire a user attribute data set and a historical behavior data set of each user, and construct each initial sample data set according to each of the knowledge attribute data sets, each of the user attribute data sets, and each of the historical behavior data sets; Update each of the initial sample data sets according to the fit scoring rule to obtain each sample data set; A natural language processing model is trained based on each of the sample data sets until the natural language processing model meets a preset training stop condition, and the natural language processing model is determined as a knowledge recommendation model.

4. The method according to claim 3, characterized in that The constructing of each initial sample data set according to each of the knowledge attribute data sets, each of the user attribute data sets and each of the historical behavior data sets includes: For each of the historical behavior data sets of the user, determining a knowledge attribute data set corresponding to each historical behavior data subset in the historical behavior data set; An initial sample data set is constructed according to the user attribute data set, the historical behavior data subset, the user behavior type and the knowledge attribute data set.

5. The method according to claim 3, characterized in that: The updating of each of the initial sample data sets according to the fit scoring rule to obtain each sample data set includes: Determine a fit score corresponding to each of the initial sample data sets according to a fit score rule; The fit score is added to the initial sample data set to obtain a sample data set.

6. The method according to claim 1, characterized in that The step of determining target recommendation knowledge from each of the initial target knowledge according to each of the degrees of fit includes: Sorting the initial target knowledge in descending order of their fit, to obtain an initial target knowledge sequence; The target recommended knowledge is determined in the initial target knowledge sequence according to a preset display quantity.

7. A knowledge recommendation device, characterized in that: The device comprises: An acquisition module is used to acquire a knowledge attribute data set of each knowledge, a target historical behavior data set of a target user, and a target user attribute data set; A screening module, used for screening each of the knowledge based on the target historical behavior data set to obtain each initial target knowledge, and constructing each knowledge user attribute data set according to the knowledge attribute data set of each initial target knowledge and the target user attribute data set; The determination module is used to determine the degree of fit between each of the initial target knowledge and the target user according to the knowledge recommendation model and each knowledge user attribute data set, and determine the target recommended knowledge in each of the initial target knowledge according to each of the degrees of fit.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images