Client behavior analysis system for love and marriage platform CRM

Abstract

The invention relates to the technical field of big data resource service, in particular to a customer behavior analysis system for a love and marriage platform CRM. The method comprises the following steps: firstly, clustering clients based on a fuzzy clustering algorithm and portrait information in a first recommendation period, and obtaining all recommended clients of a target client; further updating the membership degree of the target customer in each cluster and the recommended customer according to the interaction information of the target customer and each recommended customer in each other recommendation period; and finally, at the current moment, according to the time sequence change condition of the membership degree of the target customer in each cluster, obtaining a purpose clear index of the target customer, and further adjusting the marriage purpose of the target customer. According to the method, the interaction behavior change of the target customer and the customer recommended by the platform in different recommendation periods is analyzed, and the potential expectation of the target customer is analyzed, so that the recommended choices are continuously adjusted, the love and marriage purpose of the target customer is gradually clarified, and the matching effect of the love and marriage platform is improved.

Description

Technical Field

[0001] The present invention relates to the field of big data resource service technology, and in particular to a customer behavior analysis system for a dating platform CRM. Background Art

[0002] Customer Relationship Management (CRM) is a strategy and technology that helps businesses manage their relationships with their customers. On dating platforms, CRM systems collect and analyze basic customer profiles, behavioral information, and interaction data. This helps these platforms gain a deeper understanding of customer needs and provide personalized, refined matching or recommendation services, thereby increasing the success rate of matchmaking.

[0003] However, the user portraits in traditional marriage and love CRM systems are static and cannot adapt to the dynamic evolution of marriage and love demands. That is, when the marriage and love purpose filled in by the customer during initial registration changes over time and is not changed in time, the marriage and love platform may continue to recommend users with the same original marriage and love purpose. The contradiction between the static initial data and dynamic behavioral needs leads to inaccurate recommendations, thereby reducing the customer matching effect of the marriage and love platform. Summary of the Invention

[0004] In order to solve the technical problem of poor customer matching effect of dating platforms, the purpose of the present invention is to provide a customer behavior analysis system for dating platform CRM. The technical solution adopted is as follows:

[0005] A customer behavior analysis system for a dating platform CRM, including:

[0006] Customer information collection module: used to obtain the profile information of all customers in the CRM system of the dating platform, as well as the interaction information of any target customer with other customers during each recommendation cycle; the profile information at least includes the purpose of dating;

[0007] Intelligent customer recommendation module: used to cluster customers based on the fuzzy clustering algorithm and the portrait information in the first recommendation cycle, obtain the target customer's membership in each cluster, and obtain all recommended customers of the target customer based on the membership; in each of the remaining recommendation cycles, obtain the recommendation index of each recommended customer in the previous recommendation cycle based on the interaction information between the target customer and each recommended customer in the previous adjacent recommendation cycle, and update the target customer's membership in each cluster and the recommended customers based on the recommendation index of all recommended customers in each cluster;

[0008] Portrait dynamic update module: used to obtain the target customer's purpose clarity index based on the temporal changes in the target customer's membership in each cluster at the current moment, and adjust the target customer's marriage and love purpose in combination with the marriage and love purposes of all customers in each cluster.

[0009] Furthermore, the interaction information includes at least browsing time, number of interactions and single interaction time.

[0010] Furthermore, the method for obtaining recommended customers includes:

[0011] The membership degree of the target customer in each cluster is multiplied by a preset multiple and then rounded to the integer to obtain the recommended number of customers in each cluster, and the recommended number of customers in each cluster are screened as recommended customers of the target customer.

[0012] Furthermore, the method of selecting the recommended number of customers as recommended customers for target customers in each cluster includes:

[0013] In each cluster, customers are sorted in descending order of membership, and the recommended number of customers before sorting are taken as recommended customers.

[0014] Furthermore, the method for obtaining the recommendation index includes:

[0015] In non-first recommendation cycles, any recommended customer of the target customer in the previous recommendation cycle will be considered as the customer to be analyzed;

[0016] When the number of interactions between the target customer and the customer to be analyzed is 0, obtaining a first parameter based on the browsing time of the target customer and the customer to be analyzed, and using the first parameter as a recommendation index for obtaining the customer to be analyzed;

[0017] When the number of interactions between the target customer and the customer to be analyzed is greater than 0, a second parameter is obtained based on the number of interactions between the target customer and the customer to be analyzed and the duration of a single interaction, and the second parameter is used as a recommendation index for obtaining the customer to be analyzed; wherein the second parameter is greater than the first parameter.

[0018] Furthermore, the method for obtaining the second parameter includes:

[0019] Obtaining a total interaction duration based on the number of interactions and the duration of a single interaction, and taking the product of the total interaction duration and the number of interactions as a first interaction index;

[0020] In each non-first recommendation cycle, an interaction frequency change curve is fitted based on the total number of interactions per day between the target customer and the customer to be analyzed, and an interaction duration change curve is fitted based on the duration of each single interaction between the target customer and the customer to be analyzed; a second interaction index is obtained based on the slopes of the interaction frequency change curve and the interaction duration change curve;

[0021] The first interaction index and the second interaction index are integrated to obtain a second parameter.

[0022] Furthermore, the method of updating the membership of the target customer in each cluster and recommending customers includes:

[0023] In each cluster, the sum of the recommendation indices of all recommended customers is normalized to obtain a membership adjustment value, and the normalized value of the sum of the target customer's membership in the cluster and the membership adjustment value is used as the updated membership;

[0024] The updated membership of the target customer in each cluster is multiplied by a preset multiple and then rounded to the integer to obtain the updated recommended number of customers in each cluster. The updated recommended number of customers in each cluster are screened as the recommended customers after the target customer is updated.

[0025] Furthermore, the method for obtaining the purpose clarity index includes:

[0026] As of the current moment, the membership of the target customer in each cluster in each recommendation cycle is used as a data point to fit the time series change curve; the slope of each time series change curve is calculated, and the maximum slope is used as the comparison change parameter;

[0027] In the most recent recommendation cycle at the current moment, the cluster with the largest membership among all clusters of the target customer is selected as the target cluster, and the highlight parameter of the target cluster is obtained based on the maximum membership and its deviation from the rest of the memberships.

[0028] The slope of the time series change curve corresponding to the target cluster is used as the change parameter, and the purpose bias parameter is obtained according to the difference between the change parameter and the comparison change parameter; the highlight parameter and the purpose bias parameter are fused, and the normalized value of the fusion result is used as the purpose clarity index.

[0029] Furthermore, the method for obtaining the highlighting parameters includes:

[0030] The difference between the maximum membership and the second largest membership of the target customer in all clusters is multiplied by the maximum membership, and the product is used as the highlight parameter of the target cluster corresponding to the maximum membership.

[0031] Furthermore, the method of adjusting the target customer's marriage and love purpose includes:

[0032] When the purpose clarity index is greater than a preset threshold, the marriage and love purpose with the highest frequency among all customers in the target cluster will be used as the target customer's changed marriage and love purpose; when the purpose clarity index is less than or equal to the preset threshold, the target customer's marriage and love purpose will not be updated.

[0033] The present invention has the following beneficial effects:

[0034] The present invention first obtains the portrait information of all customers in the CRM system of the marriage and love platform, and the interaction information of any target customer with other customers in each recommendation cycle, to prepare for the analysis of subsequent customer behavior; then, in the first recommendation cycle, the customers are clustered based on the fuzzy clustering algorithm and the portrait information, and all the recommended customers of the target customer are preliminarily obtained; further, in each of the remaining recommendation cycles, the potential expectations of the target customer are analyzed based on the interaction information between the target customer and each recommended customer in the adjacent previous recommendation cycle, and the recommendation index of each recommended customer in the previous recommendation cycle is obtained, and the target customer's membership in each cluster and the recommended customers are updated according to the recommendation index of all recommended customers in each cluster, so as to continuously adjust the recommended candidates and gradually clarify the target customer's marriage and love purpose; finally, at the current moment, the target customer's tendency is clarified according to the time series change of the target customer's membership in each cluster, the target customer's purpose clarification index is obtained, and the target customer's marriage and love purpose is adjusted in combination with the marriage and love purposes of all customers in each cluster. The present invention analyzes the changes in the interactive behaviors between target customers and customers recommended by the platform in different recommendation cycles, analyzes the potential expectations of target customers, and thus continuously adjusts the recommended candidates, and then gradually clarifies the target customers' marriage and love goals, so as to improve the matching effect of the marriage and love platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0036] Figure 1 A system block diagram of a customer behavior analysis system for a dating platform CRM provided by one embodiment of the present invention;

[0037] Figure 2 A flow chart of a method for obtaining a purpose clarity index provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0038] To further illustrate the technical means and effects of the present invention to achieve the intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effects of a customer behavior analysis system for a dating platform CRM proposed by the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable form.

[0039] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0040] The following describes in detail a specific solution of a customer behavior analysis system for a dating platform CRM provided by the present invention in conjunction with the accompanying drawings.

[0041] See also Figure 1 , which shows a system block diagram of a customer behavior analysis system for a marriage and dating platform CRM provided by an embodiment of the present invention, including a customer information collection module 101, an intelligent customer recommendation module 102 and a portrait dynamic update module 103.

[0042] Customer information collection module 101: used to obtain the portrait information of all customers in the CRM system of the marriage and dating platform, and the interaction information between any target customer and other customers in each recommendation cycle; the portrait information at least includes the purpose of marriage and dating.

[0043] In one implementation of the present invention, the CRM system of the marriage and dating platform is first used to collect portrait information of each customer on the platform, where the portrait information refers to the personal information that the customer actively fills in or selects when registering on the marriage and dating platform, including: basic information such as nickname, gender, age, zodiac sign and social account; educational background such as academic qualifications, graduation school and major; professional information such as occupation and income; family information such as hometown, family members and asset information; living habits such as eating habits, work and rest time, exercise habits and smoking and drinking habits; hobbies such as music, movies, books and travel; marriage and dating purposes such as quickly finding a dating partner, quickly finding a marriage partner, ordinary making friends or long-term development, etc.; implementers can also customize the types and quantity of portrait information, such as adding ideal portrait information, but at least include marriage and dating purposes to match ideal customers with the same goals.

[0044] Taking into account that each customer's marriage and love goals may change after contacting different recommended customers, such as gradually changing from ordinary dating at the beginning to intending to fall in love or get married, the change in marriage and love goals may cause subsequent recommended customers to no longer meet their intentions; therefore, the embodiment of the present invention will set a recommendation cycle to regularly recommend matching customers to the target customer, specifically recommending customers to the target customer at the beginning of each recommendation cycle, and further analyzing the target customer's potential expectations based on the changes in the target customer's interactive behavior with the recommended customers of the marriage and love platform, so as to continuously adjust the recommended candidates, and then gradually clarify the target customer's marriage and love goals, so as to improve the matching effect of the marriage and love platform;

[0045] Therefore, in one embodiment of the present invention, each week is specifically used as a recommendation cycle. The CRM system based on the marriage and dating platform tracks the interaction information between each customer and other customers in each recommendation cycle. The interaction includes at least browsing, liking, commenting and sending messages, etc. The interaction information obtained includes the browsing time of other customers' homepages, the number of interactions with other customers and the duration of a single interaction; implementers can also customize the type and quantity of interaction information.

[0046] It should be noted that collecting customer portrait information and tracking customer behavior based on the CRM system is already an existing technology and will not be repeated here; the method of behavior analysis and portrait change for each customer on the marriage and dating platform is the same. The embodiment of the present invention will select one customer from all customers as the target customer, and analyze and describe its behavior analysis and portrait changes.

[0047] Intelligent customer recommendation module 102: used to cluster customers based on fuzzy clustering algorithm and portrait information in the first recommendation cycle, obtain the target customer's membership in each cluster, and obtain all recommended customers of the target customer based on the membership; in each of the remaining recommendation cycles, obtain the recommendation index of each recommended customer in the previous recommendation cycle based on the interaction information between the target customer and each recommended customer in the adjacent previous recommendation cycle, and update the target customer's membership and recommended customers in each cluster based on the recommendation index of all recommended customers in each cluster.

[0048] Considering that the marriage purpose reported by the target customer during the initial registration is his or her real marriage intention at that time, we can directly recommend potential customers with similar or matching profiles to him or her. Also, considering that the profile information and behavior patterns of customers on the marriage and dating platform are diverse, customers with similar profile information usually have a stronger tendency to interact and a higher success rate in matching. The fuzzy clustering algorithm can better adapt to the diversity of customers and allow a customer to belong to multiple groups, thereby helping to capture the diverse intentions and needs of customers and provide personalized recommendations for potential users.

[0049] Therefore, one embodiment of the present invention first clusters customers based on the fuzzy clustering algorithm and portrait information in the first recommendation cycle, obtains the membership of the target customer in each cluster, and then obtains all recommended customers of the target customer based on the membership. Each cluster contains similar customer groups, and the customer's membership in each cluster reflects the degree of belonging to the cluster. Recommending customers in customer groups with high membership to the target customer will improve their marriage and love matching effect.

[0050] Specifically, each personal information in the customer portrait information is first taken as a vector element to construct a multidimensional feature vector. Then, the multidimensional feature vector is used as the customer label, and the fuzzy C-means clustering algorithm is used to cluster all customers so that customers with similar labels belong to the same cluster, where the number of clusters is set to the number of types of marriage and love purposes; then the cluster containing the target customer is screened out for subsequent analysis, that is, the target customer's membership to the cluster that does not contain it is 0, and subsequent recommended customers are all customers in the cluster containing the target customer.

[0051] It should be noted that, in one embodiment of the present invention, the customers participating in fuzzy clustering are customers of the opposite sex of the target customers, and the implementer may also adjust it according to actual needs.

[0052] It should be noted that the construction of multidimensional feature vectors and the application of fuzzy C-means clustering algorithm are both existing technologies and will not be described in detail. In other embodiments, the implementer may also adopt other fuzzy clustering algorithms and may also define the number of clusters by himself.

[0053] Preferably, in one embodiment of the present invention, considering that the higher the target customer's membership relative to the cluster, the better the effect of recommending customers within the cluster to the target customer; and the membership value range is usually [0, 1], and the sum of the membership of a customer relative to all clusters is 1, the membership can be multiplied by a corresponding multiple and then rounded according to the number of recommended customers set in each recommendation cycle, thereby obtaining the number of recommended customers for each cluster, and then screening the corresponding customers for recommendation; based on this, the method for obtaining recommended customers includes:

[0054] The target customer's membership in each cluster is multiplied by a preset multiple and then rounded to the nearest integer to obtain the recommended number of customers in each cluster, and the recommended number of customers in each cluster are screened out as recommended customers for the target customer; in a preferred embodiment of the present invention, specifically in each cluster, the customers are sorted in descending order according to the membership, and the recommended number of customers before sorting are screened out as recommended customers.

[0055] As an example, the preset multiple is specifically set to 10. The preset multiple is the ideal number of recommended customers set in each recommendation cycle. The implementer can also adjust it according to the actual application situation. The membership of the target customer in each cluster is multiplied by 10 and rounded up to obtain the number of recommendations for customers in each cluster. Then, in each cluster, the recommended number of customers before the membership ranking is screened out. The union of the recommended customers in all clusters is the set of all recommended customers for the target customer.

[0056] In other examples, the implementer may also measure the similarity between the multidimensional feature vectors corresponding to each customer in each cluster and the target customer, for example, using Euclidean distance, DTW distance, etc. to measure the difference, and then sort the customers in descending order according to the difference, and then filter out a number of recommended customers who are similar to the target customer.

[0057] At this point, the customer recommendation for the target customer in the first recommendation cycle is completed; further, the interaction information between the target customer and all recommended customers can be combined to analyze and evaluate the target customer's satisfaction with each recommended customer, that is, the recommendation index, so as to analyze and evaluate the customer's intention through the customer's behavioral information, and then continuously adjust the recommended customers to clarify the target customer's marriage and love intentions.

[0058] Therefore, the embodiment of the present invention will first obtain the recommendation index of each recommended customer in the previous recommendation cycle based on the interaction information between the target customer and each recommended customer in the adjacent previous recommendation cycle in the second recommendation cycle, and further update the target customer's membership in each cluster and the recommended customers based on the recommendation index of all recommended customers in each cluster; then, by continuously repeating the above process, in each new recommendation cycle, the target customer's membership in each cluster and the recommended customers are iteratively updated to gradually clarify the target customer's marriage intention.

[0059] It should be noted that the iterative update process in each non-first recommendation cycle is consistent. Here, we only take any non-first recommendation cycle as an example for analysis and description, obtain the recommendation index of each recommended customer in the previous recommendation cycle, and update the target customer's membership and recommended customers in each cluster based on the recommendation index of all recommended customers in each cluster.

[0060] Preferably, in one embodiment of the present invention, first, any recommended customer of the target customer in the previous recommendation cycle is taken as the customer to be analyzed, and the customer to be analyzed is used as an example for analysis and description; considering that if the target customer only browses but does not interact with the customer to be analyzed, it means that the target customer has a lower interest in the customer to be analyzed, and the browsing time also indirectly reflects the interest level. The shorter the browsing time, the lower the matching intention, and the lower the recommendation index of the customer to be analyzed recommended in the previous recommendation cycle; and when the target customer interacts with the customer to be analyzed, the more interactions and the longer the interaction time, the more interested the target customer is in the customer to be analyzed, and the greater the recommendation index; but the recommendation index when interaction occurs should always be greater than the recommendation index when no interaction occurs; based on this, the method for obtaining the recommendation index includes:

[0061] When the number of interactions between the target customer and the customer to be analyzed is 0, a first parameter is obtained based on the browsing time of the target customer to the customer to be analyzed, and the first parameter is used as the recommendation index for obtaining the customer to be analyzed;

[0062] When the number of interactions between the target customer and the customer to be analyzed is greater than 0, a second parameter is obtained based on the number of interactions between the target customer and the customer to be analyzed and the duration of a single interaction, and the second parameter is used as the recommendation index for obtaining the customer to be analyzed; wherein the second parameter is greater than the first parameter.

[0063] Among them, in a preferred embodiment of the present invention, considering that the number of interactions and the total interaction duration in the recommendation cycle can help evaluate the interest of the target customer and the customer to be analyzed, a first interaction index can be obtained to help evaluate the recommendation index or the second parameter; considering that in each recommendation cycle, as the interaction deepens, the target customer and the customer to be analyzed get to know each other better, and at this time the target customer may think that the customer to be analyzed does not meet its expectations, thereby reducing the number of interactions and the interaction duration, a second interaction index can be obtained based on the change in the interaction; and then the recommendation index or the second parameter of the customer to be analyzed is evaluated by combining the two interaction indices; therefore, the method for obtaining the second parameter includes:

[0064] The total interaction time is obtained based on the number of interactions and the duration of a single interaction, and the product of the total interaction time and the number of interactions is used as the first interaction index;

[0065] In each non-first recommendation cycle, an interaction frequency change curve is fitted based on the total number of daily interactions between the target customer and the customer to be analyzed, and an interaction duration change curve is fitted based on the duration of each single interaction between the target customer and the customer to be analyzed; a second interaction index is obtained based on the slopes of the interaction frequency change curve and the interaction duration change curve;

[0066] The first interaction index and the second interaction index are integrated to obtain a second parameter.

[0067] As an example, first set a boundary between interaction and non-interaction, such as 0.3, so that the value range of the first parameter is [0, 0.3], and the value range of the second parameter is (0.3, 1];

[0068] When the number of interactions between the target customer and the customer to be analyzed is 0, the browsing time of the target customer relative to the customer to be analyzed is linearly normalized, and the normalized value is multiplied by 0.3 as the first parameter. The implementer can also directly map the first parameter to obtain the recommendation index of the customer to be analyzed.

[0069] When the number of interactions between the target customer and the customer to be analyzed is greater than 0, the duration of each interaction is accumulated to obtain the total interaction duration, and then the first interaction index is obtained; the interaction frequency change curve and the interaction duration change curve are further fitted, and the slopes of the two change curves are obtained based on the two-point formula, and the sum of the two slopes is used as the second interaction index; finally, the first interaction index and the second interaction index are added and fused, and the product is mapped to (0.3, 1] to obtain the second parameter, and then the recommendation index of the customer to be analyzed is obtained.

[0070] It should be noted that mapping, fitting of change curves and calculation of slopes are all existing technologies; in other examples, implementers can also adjust the boundaries on their own, or fuse the first interaction index and the second interaction index through basic mathematical operations such as multiplication or weighted summation, or use other normalization or mapping methods, which will not be repeated here.

[0071] By changing the customer to be analyzed, the recommendation index of each recommended customer in the previous recommendation cycle can be obtained; further considering that the higher the recommendation index of the recommended customer in each cluster, it indirectly shows that the target customer has a higher degree of intention towards the customers in the cluster, and its membership to the cluster should also be higher. Therefore, the membership of the target customer in each cluster and the recommended customers can be updated according to the recommendation index of all recommended customers in each cluster.

[0072] Preferably, in one embodiment of the present invention, considering that the recommendation index of the target customer's recommended customers in each cluster can be used as a further reference for the membership, the membership adjustment value can be first evaluated based on the recommendation indexes of all recommended customers, and then the membership of the target customer relative to the cluster can be updated and adjusted; and then the recommended customers can be updated. Based on this, the method for updating the target customer's membership in each cluster and the recommended customers includes:

[0073] In each cluster, the sum of the recommendation indices of all recommended customers is normalized to obtain the membership adjustment value. The normalized value of the sum of the target customer's membership in the cluster and the membership adjustment value is used as the updated membership;

[0074] The updated membership of the target customer in each cluster is multiplied by a preset multiple and then rounded to the integer to obtain the updated recommended number of customers in each cluster. The updated recommended number of customers in each cluster are screened as the recommended customers after the target customer is updated.

[0075] As an example, in each non-first recommendation cycle, the sum of recommendations from all recommended customers in each cluster is first linearly normalized to obtain a membership adjustment value. It should be noted that all recommended customers in each cluster participating in the recommendation index summation should not only include recommended customers selected from each cluster, but should also include recommended customers in the union of recommended customers from all clusters.

[0076] Then, the membership degree obtained by the target customer in the previous recommendation cycle is added to the sum of the membership adjustment value as the initial updated membership degree of the target customer in each cluster, and then the initial updated membership degree is normalized to obtain the updated membership degree; further, based on the same acquisition method as the recommended customers in the first recommendation cycle, the updated recommended customers of the target customer in each cluster can be obtained, and the specific steps are not repeated here.

[0077] The specific method for normalizing the initial updated membership is: taking the initial updated membership of the target customer in each cluster as the numerator, taking the sum of the initial updated membership of the target customer in all clusters as the denominator, and taking the fraction ratio as the normalized value.

[0078] At this point, the recommended customers of the target customers in each recommendation cycle can be continuously updated.

[0079] The dynamic portrait update module 103 is used to obtain the target customer's purpose clarity index according to the temporal changes of the target customer's membership in each cluster at the current moment, and adjust the target customer's marriage and love purpose in combination with the marriage and love purposes of all customers in each cluster.

[0080] Taking into account the changes in the target customer's affiliation in each cluster at the current moment, it can reflect the changes in his intention towards the cluster, determine the target customer's purpose clarity index, and thus help clarify the target customer's intention group or purpose, and then update and adjust the target customer's marriage and love purpose in combination with the marriage and love purpose of customers in the intention group.

[0081] Preferably, in one embodiment of the present invention, the method for obtaining the purpose clarity index includes:

[0082] See also Figure 2 , which shows a flow chart of a method for obtaining a purpose clarity index provided by an embodiment of the present invention, specifically comprising:

[0083] Step S201 , as of the current moment, uses the membership of the target customer in each cluster in each recommendation cycle as a data point to fit a time series change curve; calculates the slope of each time series change curve, and uses the maximum slope as a comparison change parameter.

[0084] Considering that as the recommendation cycle is updated, the target customer's membership to the cluster becomes higher, which means that the target user has a clearer intention towards the cluster and is more inclined towards this group. Fitting the time series change curve of the membership and then calculating the slope can help evaluate the change of the membership;

[0085] Therefore, as an example, taking any cluster as an example, the membership of the target customer in the cluster during each recommendation cycle is used as a data point, and then mapped to the corresponding timestamp to fit the time series change curve of the membership; then the slope of the corresponding time series change curve of each cluster is calculated based on the two-point formula, and the maximum slope is selected as the comparative change parameter; the comparative change parameter reflects the target customer's inclination towards the cluster, and also indirectly reflects the clarity of its purpose.

[0086] It should be noted that the fitting of the time series change curve and the acquisition of the slope are both existing technologies and will not be described in detail.

[0087] Step S202 : In a recommendation cycle most recent at the current moment, the cluster with the largest membership among all clusters of the target customer is selected as the target cluster, and the highlight parameter of the target cluster is obtained according to the maximum membership and its deviation from the rest of the memberships.

[0088] Considering that within the most recent recommendation cycle at the current moment, after the target customer's membership to each cluster is updated, the cluster with the largest membership will reflect the target customer's group tendency at the time of the most recent recommendation update, which can also help evaluate the target customer's purpose. Also considering that within the most recent recommendation cycle at the current moment, if the target customer's membership to a certain cluster is higher than that to other clusters, it further indicates that the target customer has a clearer intention towards that cluster, which can help evaluate the cluster's salience coefficient, and thus facilitate the subsequent comprehensive evaluation of the target customer's purpose clarity index.

[0089] In a preferred embodiment of the present invention, the cluster corresponding to the maximum membership is considered as the target cluster. When the difference between the maximum membership and the second largest membership of the target customer in all the clusters is larger, it means that the target customer has a clearer intention to match with the customer group in the target cluster, and thus the prominence coefficient of the target cluster can be evaluated. The method for obtaining the prominence parameter includes:

[0090] The difference between the maximum membership and the second largest membership of the target customer in all clusters is multiplied by the maximum membership, and the product is used as the highlight parameter of the target cluster corresponding to the maximum membership.

[0091] In step S203, the slope of the time series change curve corresponding to the target cluster is used as the change parameter, and the purpose bias parameter is obtained according to the difference between the change parameter and the comparison change parameter; the highlight parameter and the purpose bias parameter are fused, and the normalized value of the fusion result is used as the purpose clarity index.

[0092] Taking into account that the cluster corresponding to the time-series change curve corresponding to the maximum slope may not be the target cluster, but when the slope of the time-series change curve corresponding to the target cluster is closer to the maximum slope, it means that the target customer's intention to pair with customers in the target cluster is more obvious, so that the target customer's purpose bias parameter for the target cluster can be obtained, and then the purpose bias parameter and the salience parameter can be integrated to evaluate the target customer's purpose clarity index for pairing with customers in the target cluster.

[0093] As an example, the change parameter is used as the numerator, the contrast change parameter is used as the denominator, and the fraction ratio is used as the purpose bias parameter. When the ratio is closer to 1, the larger the purpose bias parameter is; then the purpose bias parameter is multiplied and fused with the highlight parameter, and the product is linearly normalized to obtain the purpose clarity index.

[0094] In other examples, the implementer may also perform a negative correlation mapping on the difference between the comparison change parameter and the change parameter to obtain the target bias parameter, such as performing an inverse operation. The larger the inverse, the larger the target bias parameter obtained. Other normalization methods may also be used, which will not be described in detail.

[0095] After obtaining the target customer's purpose clarity index, we can further evaluate the target customer's marriage and love intentions, and then remind the customer to change his or her marriage and love goals to improve the matching effect of the marriage and love platform.

[0096] Preferably, in one embodiment of the present invention, the method for adjusting the target client's love and marriage purpose includes:

[0097] When the purpose clarity index is greater than the preset threshold, the marriage and love purpose with the highest frequency among all customers in the target cluster will be used as the target customer's changed marriage and love purpose; when the purpose clarity index is less than or equal to the preset threshold, the target customer's marriage and love purpose will not be updated.

[0098] Among them, in one example of the present invention, the preset threshold is set to 0.7, and the implementer can also customize it; when the purpose clarity index is greater than 0.7, it is determined that the target customer's marriage and love purpose is gradually becoming clear, and the marriage and love platform sends a reminder to the target customer such as: "This platform analyzes your behavior and identifies that your current marriage and love purpose may deviate from your initial marriage and love purpose. We hope that you will make corrections in your personal information so that the platform can recommend more suitable matches for you"; when the target customer modifies his or her marriage and love purpose, repeat the above operation and recommend customers to him or her again until the target customer finds an ideal customer; when the purpose clarity index is less than or equal to 0.7, continue to update the recommended customers for the target customer in the new recommendation cycle until the target customer finds an ideal customer.

[0099] In summary, the present invention first obtains the portrait information of all customers and the interaction information between the target customer and other customers in each recommendation cycle; then, in the first recommendation cycle, the customers are clustered based on the fuzzy clustering algorithm and the portrait information to obtain all the recommended customers of the target customer; further, in each of the remaining recommendation cycles, the target customer's membership in each cluster and the recommended customers are updated according to the interaction information between the target customer and each recommended customer; finally, at the current moment, the target customer's purpose clarity index is obtained according to the temporal changes in the target customer's membership in each cluster, and the target customer's marriage and love purpose is adjusted. The present invention analyzes the changes in the interactive behavior of the target customer and the customers recommended by the platform in different recommendation cycles, analyzes the target customer's potential expectations, and thus continuously adjusts the recommended candidates, and then gradually clarifies the target customer's marriage and love purpose, so as to improve the matching effect of the marriage and love platform.

[0100] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0101] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. The customer behavior analysis system for the marriage and dating platform CRM is characterized by: The system comprises: Customer information collection module: used to obtain the profile information of all customers in the CRM system of the dating platform, as well as the interaction information of any target customer with other customers during each recommendation cycle; the profile information at least includes the purpose of dating; Intelligent customer recommendation module: used to cluster customers based on the fuzzy clustering algorithm and the portrait information in the first recommendation cycle, obtain the target customer's membership in each cluster, and obtain all recommended customers of the target customer based on the membership; in each of the remaining recommendation cycles, obtain the recommendation index of each recommended customer in the previous recommendation cycle based on the interaction information between the target customer and each recommended customer in the previous adjacent recommendation cycle, and update the target customer's membership in each cluster and the recommended customers based on the recommendation index of all recommended customers in each cluster; Portrait dynamic update module: used to obtain the target customer's purpose clarity index based on the temporal changes in the target customer's membership in each cluster at the current moment, and adjust the target customer's marriage and love purpose in combination with the marriage and love purposes of all customers in each cluster.

2. The customer behavior analysis system for the dating platform CRM according to claim 1 is characterized in that: The interaction information includes at least browsing time, number of interactions and single interaction time.

3. The customer behavior analysis system for the marriage and dating platform CRM according to claim 1 is characterized in that: The method for obtaining the recommended customers includes: The membership degree of the target customer in each cluster is multiplied by a preset multiple and then rounded to the integer to obtain the recommended number of customers in each cluster, and the recommended number of customers in each cluster are screened as recommended customers of the target customer.

4. The customer behavior analysis system for the dating platform CRM according to claim 3 is characterized in that: The method of selecting the recommended number of customers as recommended customers of target customers in each cluster includes: In each cluster, customers are sorted in descending order of membership, and the recommended number of customers before sorting are taken as recommended customers.

5. The customer behavior analysis system for the dating platform CRM according to claim 2 is characterized in that: The method for obtaining the recommendation index includes: In non-first recommendation cycles, any recommended customer of the target customer in the previous recommendation cycle will be considered as the customer to be analyzed; When the number of interactions between the target customer and the customer to be analyzed is 0, obtaining a first parameter based on the browsing time of the target customer and the customer to be analyzed, and using the first parameter as a recommendation index for obtaining the customer to be analyzed; When the number of interactions between the target customer and the customer to be analyzed is greater than 0, a second parameter is obtained based on the number of interactions between the target customer and the customer to be analyzed and the duration of a single interaction, and the second parameter is used as a recommendation index for obtaining the customer to be analyzed; wherein the second parameter is greater than the first parameter.

6. The customer behavior analysis system for the dating platform CRM according to claim 5 is characterized in that: The method for obtaining the second parameter includes: Obtaining a total interaction duration based on the number of interactions and the duration of a single interaction, and taking the product of the total interaction duration and the number of interactions as a first interaction index; In each non-first recommendation cycle, an interaction frequency change curve is fitted based on the total number of interactions per day between the target customer and the customer to be analyzed, and an interaction duration change curve is fitted based on the duration of each single interaction between the target customer and the customer to be analyzed; a second interaction index is obtained based on the slopes of the interaction frequency change curve and the interaction duration change curve; The first interaction index and the second interaction index are integrated to obtain a second parameter.

7. The customer behavior analysis system for the dating platform CRM according to claim 1 is characterized in that: Methods for updating the target customer's membership in each cluster and recommending customers include: In each cluster, the sum of the recommendation indices of all recommended customers is normalized to obtain a membership adjustment value, and the normalized value of the sum of the target customer's membership in the cluster and the membership adjustment value is used as the updated membership; The updated membership of the target customer in each cluster is multiplied by a preset multiple and then rounded to the integer to obtain the updated recommended number of customers in each cluster. The updated recommended number of customers in each cluster are screened as the recommended customers after the target customer is updated.

8. The customer behavior analysis system for the dating platform CRM according to claim 1 is characterized in that: The method for obtaining the purpose clarity index includes: As of the current moment, the membership of the target customer in each cluster in each recommendation cycle is used as a data point to fit the time series change curve; the slope of each time series change curve is calculated, and the maximum slope is used as the comparison change parameter; In the most recent recommendation cycle at the current moment, the cluster with the largest membership among all clusters of the target customer is selected as the target cluster, and the highlight parameter of the target cluster is obtained based on the maximum membership and its deviation from the rest of the memberships. The slope of the time series change curve corresponding to the target cluster is used as the change parameter, and the purpose bias parameter is obtained according to the difference between the change parameter and the comparison change parameter; the highlight parameter and the purpose bias parameter are fused, and the normalized value of the fusion result is used as the purpose clarity index.

9. The customer behavior analysis system for the dating platform CRM according to claim 8 is characterized in that: The method for obtaining the highlighting parameters includes: The difference between the maximum membership and the second largest membership of the target customer in all clusters is multiplied by the maximum membership, and the product is used as the highlight parameter of the target cluster corresponding to the maximum membership.

10. The customer behavior analysis system for the dating platform CRM according to claim 8 is characterized in that: The method of adjusting the target customer's marriage and love purpose includes: When the purpose clarity index is greater than a preset threshold, the marriage and love purpose with the highest frequency among all customers in the target cluster will be used as the target customer's changed marriage and love purpose; when the purpose clarity index is less than or equal to the preset threshold, the target customer's marriage and love purpose will not be updated.

Images