User recommendation method, device, equipment and storage medium based on intimacy

By obtaining and analyzing data, calculating the correlation coefficient of user indicators, determining the intimacy judgment model, and recommending users who meet the preset recommendation value, solving the problem of difficult to accurately distinguish user correlation, and improving user interaction effects and work in financial services and other fields efficiency.

CN111506826BActive Publication Date: 2024-11-19CHINA PING AN LIFE INSURANCE CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202010182282.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-03-16
Publication Date
2024-11-19
Estimated Expiration
2040-03-16

AI Technical Summary

Technical Problem

In fields such as financial business, users cannot accurately distinguish the user's relevance when selecting interactive objects, resulting in the interaction effect not achieving the goal, which reduces work efficiency.

Method used

By obtaining and analyzing source data, generating a wide table of indicator analysis, calculating user indicator correlation coefficients, determining the intimacy judgment model, and recommending users who meet the preset recommendation value.

Benefits of technology

It enhances the purpose of user recommendations and improves viscosity and work efficiency among users.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN111506826B_ABST
    Figure CN111506826B_ABST
Patent Text Reader

Abstract

The present invention discloses a user recommendation method, device, equipment and storage medium based on intimacy. The method comprises: parsing out a first-class user index and a second-class user index according to source data in a preset platform, and generating an index analysis wide table according to the first-class user index and the second-class user index; determining the index correlation coefficient between the first-class user index and the second-class user index according to a coefficient analysis model, and adding the index correlation coefficient to the index analysis wide table; obtaining multiple groups of first-class user indexes and second-class user indexes whose index correlation coefficients meet the index preselection conditions from the index analysis wide table, and determining the corresponding intimacy determination model; determining the intimacy between each first-class user and the second-class user in the second-class user association list according to all intimacy determination models, and recommending a first-class user whose intimacy meets the preset recommendation value to the second-class user. The present invention makes user recommendations based on intimacy, which enhances the purposefulness of the recommendation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of data analysis, and in particular to a user recommendation method, device, equipment and storage medium based on intimacy. Background Art

[0002] With the rapid development of the Internet, people are interacting more and more frequently in work, life and various social activities. When users choose interaction objects, they usually want to select other users with a certain degree of relevance as interaction objects for communication. However, in the actual interaction process, users often cannot distinguish whether other users are related to them from massive data (for example, there is partial overlap of user profiles, there are mutual visits or joint participation behaviors, etc.). Therefore, users often choose interaction objects randomly based on subjective consciousness, and the effect of such interaction often cannot meet user needs. For example, in financial business interactions, such as bank loans, insurance, and financial management, when users interact with salesmen, agents will select some customers with a certain degree of intimacy for communication. However, when selecting customers for communication, they can only select based on the salesmen's own subjective consciousness or experience. This selection method is not very purposeful, cannot guarantee customer conversion rate, and reduces work efficiency. Summary of the invention

[0003] The embodiments of the present invention provide an intimacy-based information recommendation method, apparatus, device and storage medium, which perform user recommendations by accurately measuring intimacy, enhance the purposefulness of recommendations, improve the stickiness between users, and improve the work efficiency of users.

[0004] A user recommendation method based on intimacy, comprising:

[0005] Get the source data in the preset platform;

[0006] Parsing the first category user index and the second category user index according to the source data, and generating an index analysis wide table according to the first category user index and the second category user index;

[0007] Determine the indicator correlation coefficient between the first category user indicator and the second category user indicator according to a preset coefficient analysis model, and add the indicator correlation coefficient to the indicator analysis wide table;

[0008] Obtaining from the indicator analysis wide table a plurality of groups of the first category user indicators and the second category user indicators whose indicator correlation coefficients satisfy the indicator pre-selection condition, and determining an intimacy determination model corresponding to each group of the first category user indicators and the second category user indicators;

[0009] The second category of users and the second category of user association list are obtained, and the intimacy between each first category of user and the second category of users in the second category of user association list is determined according to all the intimacy determination models, and the first category of users whose intimacy meets the preset recommendation value are recommended to the second category of users.

[0010] A user recommendation device based on intimacy, comprising:

[0011] A data acquisition module is used to acquire source data in a preset platform;

[0012] A parsing module, used for parsing the first category user index and the second category user index according to the source data, and generating an index analysis wide table according to the first category user index and the second category user index;

[0013] A coefficient analysis module, used to determine the indicator correlation coefficient between the first category user indicator and the second category user indicator according to a preset coefficient analysis model, and add the indicator correlation coefficient to the indicator analysis wide table;

[0014] An intimacy determination module, used to obtain from the indicator analysis wide table a plurality of groups of the first category user indicators and the second category user indicators whose indicator correlation coefficients meet the indicator pre-selection conditions, and determine an intimacy determination model corresponding to each group of the first category user indicators and the second category user indicators;

[0015] The recommendation module is used to obtain the second category of users and the second category of user association list, and determine the intimacy between each first category of users in the second category of user association list and the second category of users based on all the intimacy determination models, and recommend the first category of users whose intimacy meets the preset recommendation value to the second category of users.

[0016] A computer device includes a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor. When the processor executes the computer-readable instructions, the above-mentioned user recommendation method based on intimacy is implemented.

[0017] A computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the above-mentioned user recommendation method based on intimacy.

[0018] The user recommendation method, device, equipment and storage medium based on intimacy provided by the present invention, after generating an indicator analysis wide table using the first-class user indicators and the second-class user indicators parsed from the source data in the preset platform, obtains the indicator correlation coefficient between the first-class user indicators and the second-class user indicators, and adds it to the indicator analysis wide table, and then obtains multiple groups of first-class user indicators and second-class user indicators in the indicator analysis wide table whose indicator correlation coefficients meet the indicator pre-selection conditions, and determines the intimacy determination model corresponding to each group of first-class user indicators and second-class user indicators. At this time, the intimacy between each first-class user and the second-class user in the second-class user association list is determined according to the intimacy determination model, and the first-class user whose intimacy meets the preset recommendation value is recommended to the second-class user. The present invention accurately analyzes the commonalities between different users, that is, the degree of correlation between different indicators, through the coefficient analysis model, and measures the intimacy between different users according to the commonalities between different users, and then makes user recommendations based on intimacy, which enhances the purpose of recommendation, improves the viscosity between users, and improves the work efficiency of users. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0020] Figure 1 is a schematic diagram of an application environment of a user recommendation method based on intimacy in one embodiment of the present invention;

[0021] Figure 2 is a flow chart of a user recommendation method based on intimacy in one embodiment of the present invention;

[0022] Figure 3 is a flowchart of step S10 of a user recommendation method based on intimacy in one embodiment of the present invention;

[0023] Figure 4 is a flowchart of step S30 of the user recommendation method based on intimacy in one embodiment of the present invention;

[0024] Figure 5 is a flowchart of step S40 of the user recommendation method based on intimacy in one embodiment of the present invention;

[0025] Figure 6 is a principle block diagram of a user recommendation device based on intimacy in one embodiment of the present invention;

[0026] Figure 7is a principle block diagram of an intimacy determination module of an intimacy-based user recommendation device in one embodiment of the present invention;

[0027] Figure 8 is a schematic diagram of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0029] The user recommendation method based on intimacy provided by the present invention can be applied in the following aspects: Figure 1 In an application environment, a client communicates with a server via a network. The client includes, but is not limited to, various personal computers, laptops, smart phones, tablet computers, cameras, and portable wearable devices. The server can be implemented as an independent server or a server cluster consisting of multiple servers.

[0030] In one embodiment, if Figure 2 As shown, a user recommendation method based on intimacy is provided, and the method is applied in Figure 1 The server in the example is used to illustrate the following steps:

[0031] S10, obtaining source data in a preset platform.

[0032] In this embodiment, the source data refers to data used to analyze the intimacy between a class of users and a class of users in a preset platform, including but not limited to the information of a class of users, a class of users, and the behavioral information of a class of users and a class of users in the preset platform. Among them, the behavioral information of a class of users and a class of users includes but is not limited to information such as mutual visit behavior, one-way visit behavior, and joint participation. The class of users and the class of users can be set according to demand. For example, if the preset platform is a financial business service platform set up for financial business, the class of users can be customers registered on the financial business service platform, and the class of users can be salesmen. For another example, if the preset platform is a game platform, the class of users can be game users who have not formed a game team on the game platform, and the class of users refers to game users who have formed a game team on the game platform.

[0033] Preferably, in a specific embodiment, Figure 2 As shown, the step S10 specifically includes the following steps:

[0034] S101, receiving a data collection instruction, and starting a first collection thread, a second collection thread, and a third collection thread according to the data collection instruction.

[0035] That is, when a data collection instruction sent by a user is received, a first collection thread for collecting information of a type of user, a second collection thread for collecting information of a type of user, and a third collection thread for collecting behavioral information of the type of users and the type of users are started, so as to collect data from the preset platform through different collection threads, which is conducive to improving data collection efficiency.

[0036] S102: Collect a type of user information in the preset platform through the first collection thread.

[0037] Taking a type of user as a customer who registers on a preset platform as an example, the customer's occupational level, educational background and other information are collected through the first collection thread as a type of user information.

[0038] S103: Collect the second type of user information in the preset platform through the second collection thread.

[0039] Taking the second-category user as a salesperson as an example, the salesperson's company length of service, work experience, work performance and other information are collected through the second collection thread as the second-category user information.

[0040] S104: Collecting behavior information of the first category of users and the second category of users in the preset platform through the third collection thread.

[0041] Taking the first category of users as customers and the second category of users as salesmen as examples, the number of times customers visit salesmen (or salesmen visit customers) within a period of time is collected through the third collection thread as mutual visit behavior, the number of times customers comment, reply, like, forward, etc. on the content posted by the salesmen on the preset platform within a period of time is collected as one-way visit behavior, and the number of times customers and salesmen jointly participate in offline interactive visits, product buying and selling, etc. is collected as joint participation, and information such as mutual visit behavior, one-way visit behavior and joint participation is used as behavioral information of the first category of users and the second category of users.

[0042] S105: Obtain source data according to the first category of user information, the second category of user information, and the behavior information.

[0043] That is, the first-category user information collected by the first collection thread, the second-category user information collected by the second collection thread, and the behavior information of the first-category user and the second-category user collected by the third collection thread are stored in the database as source data in the preset platform.

[0044] It is understandable that the step S102, the step S103 and the step S104 may be performed with priority given to one of them, or may be performed simultaneously.

[0045] S20, parsing the source data to obtain a first-category user index and a second-category user index, and generating an index analysis wide table according to the first-category user index and the second-category user index.

[0046] Specifically, the first keyword and the second keyword are set according to the needs, the first text containing the first keyword is extracted from the source data, and the probability of the first keyword in the first text is counted, and then the first user index is determined according to the probability of the first keyword, that is, the first keyword whose probability of appearance in the first text meets the preset probability threshold is used as the first user index; at the same time, the second text containing the second keyword is extracted from the source data, and the probability of the second keyword in the second text is counted, and then the second user index is determined according to the probability of the second keyword. Taking the first user as a customer and the second user as a salesperson as an example, the first user index can be the number of times the customer clicks or forwards the article published by the salesperson in one month, and the second user index can be the number of customers owned by the salesperson.

[0047] Further, an indicator analysis wide table is generated based on the first category user indicator and the second category user indicator. p The row is used to represent the pth first-class user index, which analyzes the Nth q The column is used to represent the qth second-class user index, and the Mth p Row and Nth q The corresponding position of the column can represent the correlation degree between the p-th first-category user indicator and the q-th second-category user indicator.

[0048] Preferably, in a specific embodiment, step S20 specifically includes the following steps:

[0049] First, a type of user index is extracted from the source data according to a preset first keyword, and a type of user index is extracted from the source data according to a preset second keyword; then, the index analysis wide table is constructed with the type of user index as rows and the type of user index as columns.

[0050] Specifically, a first text containing a first keyword and a second text containing a second keyword are respectively extracted from the source data, a first user indicator is determined according to the occurrence probability of the first keyword in the first text, and two types of user indicators are determined according to the occurrence probability of the first keyword in the second text, and then the number of the first type of user indicators and the number of the second type of user indicators are respectively counted, and an M-row and N-column indicator analysis wide table is constructed with the first type of user indicators as rows and the second type of user indicators as columns, wherein M is the number of rows of the indicator analysis wide table, and N is the number of columns of the indicator analysis wide table, and the number of rows of the indicator analysis wide table is the same as the number of the first type of user indicators, and the number of columns of the indicator analysis wide table is the same as the number of the second type of user indicators.

[0051] S30, determining the indicator correlation coefficient between the first category user indicator and the second category user indicator according to a preset coefficient analysis model, and adding the indicator correlation coefficient to the indicator analysis wide table.

[0052] Preferably, in a specific embodiment, step S30 specifically includes the following steps:

[0053] S301, selecting the first category user index and the second category user index from the index analysis wide table.

[0054] S302, respectively obtaining a first analysis sample corresponding to the first category of user indicators and a second analysis sample corresponding to the second category of user indicators; the first analysis sample includes multiple indicator values ​​associated with the first category of user indicators; the second analysis sample includes multiple indicator values ​​associated with the second category of user indicators.

[0055] S303: Input the first analysis sample and the second analysis sample into a preset coefficient analysis model, and receive the indicator correlation coefficient output by the coefficient analysis model.

[0056] S304, adding the indicator correlation coefficient to a corresponding position in the indicator analysis wide table; the corresponding position is determined by the row where the selected indicator of the first category of users is located and the column where the selected indicator of the second category of users is located.

[0057] Specifically, first obtain a first analysis sample corresponding to a first category of user indicators and a second analysis sample corresponding to a second category of user indicators, wherein the first analysis sample contains multiple indicator values ​​associated with the first category of user indicators (for example, there are N agents in the preset platform, and the length of service of each agent is S1, S2, ..., Sn, respectively, then the first analysis sample is a sample containing S1, S2, ..., Sn), and the second analysis sample contains multiple indicator values ​​associated with the second category of user indicators. Preferably, the data volume of the first analysis sample is the same as the data volume of the second analysis sample; then input the first analysis sample and the second analysis sample into a preset coefficient analysis model, and let the coefficient analysis model automatically calculate the indicator correlation coefficient between the first category of user indicators and the second category of user indicators according to all indicator values ​​contained in the first analysis sample and all indicator values ​​contained in the second analysis sample, and output the calculated indicator correlation coefficient; finally, when receiving the indicator correlation coefficient output by the coefficient analysis model, determine the specific position of the indicator correlation coefficient in the indicator analysis wide table according to the row where the first category of user indicators are located and the column where the second category of user indicators are located, and add the indicator correlation coefficient to the determined specific position. Preferably, the coefficient analysis model is:

[0058]

[0059] Among them, r is the index correlation coefficient; X i is the i-th index value in the first analysis sample corresponding to the one type of user index; Y i is the i-th indicator value in the second analysis sample corresponding to the second type of user indicators; n is the number of indicator values.

[0060] It can be understood that the index correlation coefficient can measure the correlation between the first category user index and the second category user index, and the value range of the index correlation coefficient is [-1, 1]. If the absolute value of the index correlation coefficient is larger, it indicates that the correlation between the first category user index and the second category user index is higher; and if the first category user index and the second category user index are linearly correlated, the index correlation coefficient takes a value of 1 (positive linear correlation) or -1 (negative linear correlation).

[0061] S40, obtaining from the indicator analysis wide table a plurality of groups of the first category user indicators and the second category user indicators whose indicator correlation coefficients satisfy the indicator pre-selection conditions, and determining an intimacy determination model corresponding to each group of the first category user indicators and the second category user indicators.

[0062] In this embodiment, the indicator pre-selection condition is that the absolute value of the indicator correlation coefficient between the first category user indicator and the second category user indicator is greater than or equal to a preset coefficient threshold, for example, 0.5.

[0063] The intimacy determination model is composed of a plurality of association rules; the association rules are generated according to a group of first-class user indicators and second-class user indicators whose indicator correlation coefficients meet the indicator pre-selection conditions, and the association rules are implication formulas in the form of "A→B", where A is called the prerequisite of the association rule and B is called the inevitable result of the association rule. Preferably, a group of first-class user indicators and second-class user indicators are associated with only one intimacy determination model.

[0064] Specifically, at least one group of first-class user indicators and second-class user indicators whose absolute value of the indicator correlation coefficient in the indicator analysis wide table is greater than or equal to the preset coefficient threshold is marked, that is, if the absolute value of the indicator correlation coefficient between the p-th first-class user indicator and the q-th second-class user indicator in the indicator analysis wide table is greater than or equal to the preset coefficient threshold, then the p-th first-class user indicator and the q-th second-class user indicator are marked in the indicator analysis wide table. At this time, the job record information in the preset platform can be used to extract the first information sample of the p-th first-class user indicator from the job record information (the first information sample may have multiple identical indicator values ​​associated with the p-th first-class user indicator), and generate a first set based on the first information sample, and extract the second information sample of the q-th second-class user indicator from the job record information (the first information sample may have multiple identical indicator values ​​associated with the q-th second-class user indicator), and generate a second set based on the second information sample; wherein the first set and the second set contain at least one item.

[0065] Furthermore, a job set including multiple item sets is generated based on all job record information in the preset platform, and the first set, the second set and the job set are input into the preset intimacy analysis model, and the intimacy determination model output by the intimacy analysis model is received, and the indicator correlation coefficient between the p-th first-class user indicator and the q-th second-class user indicator is associated with the intimacy determination model and stored. Preferably, the intimacy analysis model is the intimacy analysis model in step S404, which includes rule preselection conditions and a calculation model for calculating the support, confidence and lift of the rule.

[0066] S50, obtaining the second category of users and the second category of user association list, and determining the intimacy between each first category of users in the second category of user association list and the second category of users according to all the intimacy determination models, and recommending the first category of users whose intimacy meets a preset recommendation value to the second category of users according to the intimacy.

[0067] It is understandable that when a recommendation request is received from a second-category user in the preset platform, the first-category user associated with the second-category user is obtained according to the recommendation request, and the second-category user association list is generated according to the first-category user associated with the second-category user. At this time, the intimacy between each first-category user in the second-category user association list and the second-category user is determined according to the intimacy determination model corresponding to each group of first-category user indicators and second-category user indicators obtained in step S40, and then the first-category user whose matching degree meets the preset recommendation value is recommended to the second-category user according to the intimacy.

[0068] Preferably, the step S50 specifically includes the following steps:

[0069] First, a recommendation request sent by a second category of users is received, and a second category of users association list is obtained according to the recommendation request, wherein the second category of users association list includes a first category of users associated with the second category of users.

[0070] Then, the first category user index of each first category user in the second category user association list is obtained, and the first category user index of the first category user is obtained.

[0071] Finally, based on the first-category user indicators of each first-category user in the second-category user association list, the second-category user indicators of the second-category users, and each of the intimacy judgment models, the intimacy between each first-category user and the second-category user is obtained, and all first-category users whose intimacy meets the preset recommended value are sent to the second-category users as a recommended user list.

[0072] It can be understood that after receiving the recommendation request sent by the second-class user, a first-class user is randomly selected from the second-class user association list, and all the first-class user indicators of the first-class user and the indicator value associated with each first-class user indicator are obtained. At the same time, all the second-class user indicators of the second-class user and the indicator value associated with each second-class user indicator are obtained. At this time, all the first-class user indicators of the first-class user and all the second-class user indicators of the second-class user are marked as primary query information. At the same time, multiple array sets are generated according to the indicator values ​​associated with all the first-class user indicators and the indicator values ​​associated with all the second-class user indicators, and the generated array sets are marked as secondary query information.

[0073] Further, the index correlation coefficient that satisfies the first query information is queried from the database, the intimacy determination model associated with the index correlation coefficient is marked, and the intimacy determination model that satisfies the second query information is queried from the marked intimacy determination model, that is, whether there is an association rule that satisfies the second query information in the marked intimacy determination model is detected. When there is an association rule that satisfies the second query information in the marked intimacy determination model, the intimacy determination model is determined to be the intimacy determination model that satisfies the second query information, the number of rule matches is obtained by counting the intimacy determination models that satisfy the second query information, the total number of rules contained in all intimacy determination models is accumulated, and the intimacy between the first category of users and the second category of users is obtained based on the number of rule matches and the total number of rules. After obtaining the intimacy between each first category of users and the second category of users in the second category of user association list according to the above-mentioned intimacy calculation method, all first category users whose intimacy meets the preset recommended value are sent to the second category of users as a recommended user list.

[0074] In summary, this embodiment uses the source data in the preset platform to parse out the first-class user indicators and the second-class user indicators to generate the indicator analysis wide table, and then determines the indicator correlation coefficient between the first-class user indicators and the second-class user indicators according to the preset coefficient analysis model, and adds it to the indicator analysis wide table, and then obtains multiple groups of first-class user indicators and second-class user indicators in the indicator analysis wide table whose indicator correlation coefficients meet the indicator pre-selection conditions, and determines each group of first-class user indicators and second-class user indicators to obtain the corresponding intimacy determination model. At this time, the intimacy between each first-class user and the second-class user in the second-class user association list is determined according to the intimacy determination model, and the first-class users whose intimacy meets the preset recommendation value are recommended to the second-class users. This embodiment accurately analyzes the commonalities between different users through the coefficient analysis model, and then measures the intimacy between different users according to the commonalities between different users, and then makes user recommendations based on intimacy, which enhances the purpose of recommendation, improves the stickiness between users, and improves the work efficiency of users.

[0075] In one embodiment, if Figure 5 As shown, the step S40 specifically includes the following steps:

[0076] S401, obtaining job record information in the preset platform, and generating a job set including multiple item sets according to the job record information; each piece of the job record information corresponds to one of the item sets in the job set.

[0077] In this embodiment, each piece of the job record information may include information such as a first-class user indicator and an indicator value associated with the first-class user indicator, a second-class user indicator, and an indicator value associated with the second-class user indicator.

[0078] Specifically, all job record information during a preset collection period (for example, the first day to the last day of a quarter is a collection period) in the preset platform is obtained, and a group of corresponding indicator values ​​are extracted from each job record information, and then an item set is formed according to the extracted group of indicator values, and a job set is formed according to all item sets.

[0079] S402, marking the first category of user indicators and the second category of user indicators whose indicator correlation coefficients in the indicator analysis wide table meet the indicator pre-selection conditions, and extracting a first information sample corresponding to the marked first category of user indicators and a second information sample corresponding to the marked second category of user indicators from the job record information.

[0080] It can be understood that the wide index analysis table includes multiple first-class user indexes of the first-class user, multiple second-class user indexes of the second-class user, and the index correlation coefficient between each first-class user index and each second-class user index. That is, when the wide index analysis table includes M first-class user indexes and N second-class user indexes, the index correlation coefficients in the wide index analysis table are M*N.

[0081] Preferably, all the job record information obtained in step S401 is stored in a job record file, and the priority order of the first category of user indicators and the priority order of the second category of user indicators can be set according to the needs, and after all the job record information in the job record file is sorted according to the priority order of the first category of user indicators, a first backup file is obtained, and after all the job record information in the job record file is sorted according to the priority order of the second category of user indicators, a second backup file is obtained. At this time, after determining the first category of user indicators and the second category of user indicators whose absolute values ​​of the relevant indicator coefficients are greater than or equal to the preset coefficient threshold, the first information sample corresponding to the first category of user indicators is extracted based on the first backup file, and the second information sample corresponding to the second category of user indicators is extracted based on the second backup file, which is conducive to improving data processing efficiency.

[0082] It should be noted that for the first-class user indicators and second-class user indicators whose absolute values ​​of the relevant indicator coefficients are less than the preset coefficient thresholds, the two do not have an influence relationship, and the extracted association rules are of little significance and cannot contribute to the overall intimacy association rules.

[0083] S403: Generate a first set corresponding to the first type of user indicators according to the first information sample, and generate a second set corresponding to the second type of user indicators according to the second information sample.

[0084] Specifically, multiple indicator values ​​associated with a type of user indicators contained in the first information sample are merged to obtain a first set, and multiple indicator values ​​associated with two types of user indicators contained in the second information sample are merged to obtain a second set, and each element in the first set is different from that in the first set.

[0085] S404: Input the first set, the second set and the operation set into a preset intimacy analysis model, and receive and store an intimacy determination model output by the intimacy analysis model.

[0086] In this embodiment, after the first set, the second set and the operation set are input into the intimacy analysis model, the intimacy analysis model will obtain a combination set based on the first set and the second set (each combination set corresponds to an association rule), and automatically calculate the support, confidence and lift of the rule based on the combination set and the operation set, and then automatically screen out the association rules that meet the rule pre-selection conditions based on the support, confidence and lift of the rule, and generate an intimacy determination model based on the association rules that meet the rule pre-selection conditions and then output it. Preferably, the rule pre-selection conditions and the calculation model for calculating the support, confidence and lift of the rule have been pre-configured in the intimacy analysis model. Among them, the rule pre-selection conditions refer to that the lift of the rule is greater than the preset lift threshold, and the confidence of the rule satisfies the highest confidence, and the support of the rule satisfies the minimum support under the premise of satisfying the above two sub-conditions.

[0087] The support can represent the probability of a combination set appearing in a job set. If the job set is I = {i1, i2, ..., in}, the first set is X = {A1, A2, ..., An}, the second set is Y = {B1, B2, ..., Bm}, in is the item set in the job set, A is an element in the first set, and B is an element in the second set, then the calculation model of the support is:

[0088] Support(A→B)=P(A,B) / P(I)=P(A∪B) / P(I)=num(A∪B) / num(I)

[0089] Among them, Support(A→B) represents the support of the association rule "A→B"; num(A∪B) represents the number of combinations of elements in set X and set Y that appear in the job set (the number is also called the frequency); num(I) represents the number of all item sets in the job set.

[0090] The confidence can represent the probability of the occurrence of the inevitable result B deduced from the association rule "A→B" when the prerequisite A occurs, that is, the possibility of the combination set containing element A containing element B. The calculation model of the confidence can be:

[0091] Confidence(A→B)=P(B|A)=P(A∪B) / P(A)

[0092] Among them, Confidence(A→B) represents the confidence of the association rule “A→B”; P(B|A) represents the probability of the occurrence of the inevitable result B inferred from the association rule “A→B” when the prerequisite A occurs.

[0093] The lift can represent the ratio of the probability of the necessary result Y under the condition containing the prerequisite A to the probability of the total occurrence of B. The calculation model of the lift can be:

[0094] Lift(A→B)=P(B|A) / P(B)

[0095] Among them, Lift(A→B) represents the lift of the association rule “A→B”; P(B) represents the total probability of B appearing.

[0096] In summary, this embodiment obtains the job record information in the preset platform, generates a job set according to the job record information, extracts the first information sample corresponding to the marked first category of user indicators from the job record information, and generates the first set according to the first information sample, and extracts the second information sample corresponding to the marked second category of user indicators from the job record information, and generates the second set according to the second information sample, and then inputs the first set, the second set and the job set into the intimacy analysis model, and receives the intimacy determination model output by the intimacy analysis model. This embodiment uses the intimacy analysis model to obtain the intimacy determination model, and then obtains the intimacy between the first category of users and the second category of users according to the intimacy determination model, thereby improving the accuracy of the intimacy.

[0097] In one embodiment, if Figure 6 As shown, a user recommendation device based on intimacy is provided, and the user recommendation device based on intimacy corresponds one-to-one with the user recommendation method based on intimacy in the above embodiment. The user recommendation device based on intimacy includes the following modules, and each functional module is described in detail as follows:

[0098] The data acquisition module 110 is used to acquire source data in a preset platform.

[0099] The parsing module 120 is used to parse the source data to obtain the first category user index and the second category user index, and generate an index analysis wide table according to the first category user index and the second category user index.

[0100] The coefficient analysis module 130 is used to determine the indicator correlation coefficient between the first category user indicator and the second category user indicator according to a preset coefficient analysis model, and add the indicator correlation coefficient to the indicator analysis wide table.

[0101] The intimacy determination module 140 is used to obtain multiple groups of the first category user indicators and the second category user indicators whose indicator correlation coefficients meet the indicator pre-selection conditions from the indicator analysis wide table, and determine the intimacy determination model corresponding to each group of the first category user indicators and the second category user indicators.

[0102] The recommendation module 150 is used to obtain the second category users and the second category user association list, and determine the intimacy between each first category user in the second category user association list and the second category user according to all the intimacy determination models, and recommend the first category users whose intimacy meets the preset recommendation value to the second category users.

[0103] In one embodiment, the data acquisition module 110 includes the following submodules, and each functional submodule is described in detail as follows:

[0104] The receiving submodule is used to receive a data collection instruction and start the first collection thread, the second collection thread and the third collection thread according to the data collection instruction.

[0105] The first collection submodule is used to collect a type of user information in the preset platform through the first collection thread.

[0106] The second collection submodule is used to collect the second type of user information in the preset platform through the second collection thread.

[0107] The third collection submodule is used to collect the behavior information of the first category of users and the second category of users in the preset platform through the third collection thread.

[0108] The data combination submodule is used to obtain source data according to the first type of user information, the second type of user information and the behavior information.

[0109] In one embodiment, the parsing module 120 includes the following submodules, and each functional submodule is described in detail as follows:

[0110] The extraction submodule is used to extract a type of user indicators from the source data according to a preset first keyword, and to extract a type of user indicators from the source data according to a preset second keyword.

[0111] A construction submodule is used to construct the indicator analysis wide table using the first type of user indicators as rows and the second type of user indicators as columns.

[0112] In one embodiment, the coefficient analysis module 130 includes the following submodules, and each functional submodule is described in detail as follows:

[0113] The selection submodule is used to select the first category user index and the second category user index from the index analysis wide table.

[0114] The sample acquisition submodule is used to respectively acquire a first analysis sample corresponding to the first category of user indicators and a second analysis sample corresponding to the second category of user indicators; the first analysis sample includes multiple indicator values ​​associated with the first category of user indicators; the second analysis sample includes multiple indicator values ​​associated with the second category of user indicators.

[0115] The coefficient analysis submodule is used to input the first analysis sample and the second analysis sample into a preset coefficient analysis model, and receive the indicator correlation coefficient output by the coefficient analysis model.

[0116] The adding submodule is used to add the indicator correlation coefficient to the corresponding position in the indicator analysis wide table; the corresponding position is determined by the row where the selected first category user indicator is located and the column where the selected second category user indicator is located.

[0117] In one embodiment, in the coefficient analysis module 130, the coefficient analysis model is:

[0118]

[0119] Among them, r is the index correlation coefficient; X i is the i-th index value in the first analysis sample corresponding to the one type of user index; i is the i-th indicator value in the second analysis sample corresponding to the second type of user indicators; n is the number of indicator values.

[0120] In one embodiment, if Figure 7 As shown, the intimacy determination module 140 includes the following submodules, and each functional submodule is described in detail as follows:

[0121] The information acquisition submodule 141 is used to acquire the job record information in the preset platform, and generate a job set including multiple item sets according to the job record information; each piece of the job record information corresponds to one of the item sets in the job set.

[0122] The sample extraction submodule 142 is used to mark the first category of user indicators and the second category of user indicators whose indicator correlation coefficients in the indicator analysis wide table meet the indicator pre-selection conditions, and extract the first information sample corresponding to the marked first category of user indicators and the second information sample corresponding to the marked second category of user indicators from the job record information.

[0123] The set generation submodule 143 is used to generate a first set corresponding to the first type of user indicators according to the first information sample, and to generate a second set corresponding to the second type of user indicators according to the second information sample.

[0124] The output submodule 144 is used to input the first set, the second set and the operation set into a preset intimacy analysis model, and receive and store the intimacy determination model output by the intimacy analysis model.

[0125] For the specific definition of the user recommendation device based on intimacy, please refer to the definition of the user recommendation method based on intimacy above, which will not be repeated here. Each module in the above-mentioned user recommendation device based on intimacy can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0126] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 8 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. 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, computer-readable instructions and a database. The internal memory provides an environment for the operation of the operating system and computer-readable instructions in the non-volatile storage medium. When the computer-readable instructions are executed by the processor, a user recommendation method based on intimacy is implemented.

[0127] In one embodiment, a computer device is provided, including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, and when the processor executes the computer-readable instructions, the steps of the user recommendation method based on intimacy in the above embodiment are implemented.

[0128] In one embodiment, a computer-readable storage medium is provided, on which computer-readable instructions are stored. When the computer-readable instructions are executed by a processor, the steps of the user recommendation method based on intimacy in the above embodiment are implemented.

[0129] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a non-volatile computer-readable storage medium. When the computer-readable instructions are executed, they can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0130] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units or modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional units or modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0131] The embodiments described above 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 may be replaced by equivalents. Such 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, and should all be included in the protection scope of the present invention.

Claims

1. A user recommendation method based on intimacy, It is characterized in that include: Get the source data in the preset platform; Parsing the first category user index and the second category user index according to the source data, and generating an index analysis wide table according to the first category user index and the second category user index; Determine the indicator correlation coefficient between the first category user indicator and the second category user indicator according to a preset coefficient analysis model, and add the indicator correlation coefficient to the indicator analysis wide table; Obtaining from the indicator analysis wide table a plurality of groups of the first category user indicators and the second category user indicators whose indicator correlation coefficients satisfy the indicator pre-selection condition, and determining an intimacy determination model corresponding to each group of the first category user indicators and the second category user indicators; Obtaining the second category of users and the second category of user association list, and determining the intimacy between each first category of users in the second category of user association list and the second category of users according to all the intimacy determination models, and recommending the first category of users whose intimacy meets a preset recommendation value to the second category of users; The step of obtaining multiple groups of the first category user indicators and the second category user indicators whose indicator correlation coefficients satisfy the indicator pre-selection conditions from the indicator analysis wide table, and determining the intimacy determination model corresponding to each group of the first category user indicators and the second category user indicators, includes: Acquire the job record information in the preset platform, and generate a job set including a plurality of item sets according to the job record information; each piece of the job record information corresponds to one of the item sets in the job set; Marking the first category of user indicators and the second category of user indicators whose indicator correlation coefficients in the indicator analysis wide table meet the indicator pre-selection condition, and extracting a first information sample corresponding to the marked first category of user indicators and a second information sample corresponding to the marked second category of user indicators from the job record information; Generate a first set corresponding to the first type of user indicators according to the first information sample, and generate a second set corresponding to the second type of user indicators according to the second information sample; Inputting the first set, the second set and the operation set into a preset intimacy analysis model, and receiving and storing an intimacy determination model output by the intimacy analysis model; The step of parsing the first category user index and the second category user index according to the source data, and generating an index analysis wide table according to the first category user index and the second category user index, includes: Extracting a type of user index from the source data according to a preset first keyword, and extracting a second type of user index from the source data according to a preset second keyword; The indicator analysis wide table is constructed with the first type of user indicators as rows and the second type of user indicators as columns.

2. The user recommendation method based on intimacy as claimed in claim 1, It is characterized in that The obtaining of source data in the preset platform includes: Receive a data collection instruction, and start a first collection thread, a second collection thread, and a third collection thread according to the data collection instruction; Collecting a type of user information in the preset platform through the first collection thread; Collecting the second type of user information in the preset platform through the second collection thread; Collecting the behavior information of the first category of users and the second category of users in the preset platform through the third collection thread; The source data is obtained according to the first category of user information, the second category of user information and the behavior information.

3. The user recommendation method based on intimacy as claimed in claim 1, It is characterized in that The determining the indicator correlation coefficient between the first category user indicator and the second category user indicator according to a preset coefficient analysis model, and adding the indicator correlation coefficient to the indicator analysis wide table, includes: Selecting the first category user index and the second category user index from the index analysis wide table; Respectively obtaining a first analysis sample corresponding to the first type of user indicators and a second analysis sample corresponding to the second type of user indicators; the first analysis sample includes multiple indicator values ​​associated with the first type of user indicators; the second analysis sample includes multiple indicator values ​​associated with the second type of user indicators; Inputting the first analysis sample and the second analysis sample into a preset coefficient analysis model, and receiving the indicator correlation coefficient output by the coefficient analysis model; The indicator correlation coefficient is added to a corresponding position in the indicator analysis wide table; the corresponding position is determined by the row where the selected indicator of the first category of users is located and the column where the selected indicator of the second category of users is located.

4. The user recommendation method based on intimacy as claimed in claim 3, It is characterized in that The coefficient analysis model is: in, is the indicator correlation coefficient; is the first analysis sample corresponding to the user index of the first type. Index value; is the first one in the second analysis sample corresponding to the second type of user indicators Index value; is the number of indicator values.

5. A user recommendation device based on intimacy, It is characterized in that include: A data acquisition module is used to acquire source data in a preset platform; A parsing module, used for parsing the first category user index and the second category user index according to the source data, and generating an index analysis wide table according to the first category user index and the second category user index; A coefficient analysis module, used to determine the indicator correlation coefficient between the first category user indicator and the second category user indicator according to a preset coefficient analysis model, and add the indicator correlation coefficient to the indicator analysis wide table; An intimacy determination module, used to obtain from the indicator analysis wide table a plurality of groups of the first category user indicators and the second category user indicators whose indicator correlation coefficients meet the indicator pre-selection conditions, and determine an intimacy determination model corresponding to each group of the first category user indicators and the second category user indicators; A recommendation module, used to obtain the second category of users and the second category of user association list, and determine the intimacy between each first category of users in the second category of user association list and the second category of users according to all the intimacy determination models, and recommend the first category of users whose intimacy meets the preset recommendation value to the second category of users; The intimacy determination module comprises: An information acquisition submodule, used to acquire job record information in the preset platform, and generate a job set including multiple item sets according to the job record information; each piece of the job record information corresponds to one item set in the job set; A sample extraction submodule, used for marking the first category of user indicators and the second category of user indicators whose indicator correlation coefficients in the indicator analysis wide table meet the indicator pre-selection condition, and extracting a first information sample corresponding to the marked first category of user indicators and a second information sample corresponding to the marked second category of user indicators from the job record information; A set generation submodule, configured to generate a first set corresponding to the first type of user indicators according to the first information sample, and to generate a second set corresponding to the second type of user indicators according to the second information sample; an output submodule, configured to input the first set, the second set and the operation set into a preset intimacy analysis model, and receive and store an intimacy determination model output by the intimacy analysis model; The parsing module comprises: An extraction submodule, configured to extract a first type of user indicator from the source data according to a preset first keyword, and to extract a second type of user indicator from the source data according to a preset second keyword; A construction submodule is used to construct the indicator analysis wide table using the first type of user indicators as rows and the second type of user indicators as columns.

6. A computer device comprising a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, It is characterized in that When the processor executes the computer-readable instructions, the intimacy-based user recommendation method according to any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium storing computer-readable instructions. It is characterized in that When the computer-readable instructions are executed by a processor, the user recommendation method based on intimacy as described in any one of claims 1 to 4 is implemented.

Citation Information

Patent Citations

  • Node-closeness-based social network site friend recommendation method

    CN103823888A

  • Information recommendation method and device based on intimacy and computer readable storage medium

    CN109996192A

  • Expression recommendation method and device based on user social relation

    CN110019286A

  • Social network service system and method for recommending friend of friend based on intimacy between users

    US20130166574A1