Sales clue identification method and device based on knowledge graph

Abstract

The invention relates to a sales clue identification method and device based on a knowledge graph, and the method comprises the steps: collecting the user feature information of a target user and the associated user feature information of a plurality of associated users, and carrying out the retrieval in a sales user database constructed based on the knowledge graph based on a plurality of association relationship categories, obtaining a matched purchased user feature information set and a plurality of matched associated user feature information sets; and according to the matched purchased user feature information set and the plurality of matched associated user feature information sets, performing clue analysis on the user feature information and the plurality of associated user feature information to obtain the clue degree of the target user, and generating a sales clue identification result. According to the method and the device, the technical problems of relatively low sales clue identification efficiency and poor accuracy of the identified sales clues in the prior art are solved by automatically identifying the sales clues by using the algorithm.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular, to a method and device for identifying sales leads based on a knowledge graph. Background Art

[0002] In the sales process, mining sales leads is a crucial link. Whether sales leads can be efficiently identified and the quality of the identified sales leads directly affect the sales performance and are related to the survival of the enterprise. However, traditional sales lead identification methods mostly rely on advertising recommendations based on user access records, collecting information of participants at industry exhibitions or events, sending email invitations to potential users, and then manually identifying and judging valuable customers. There are technical problems such as low efficiency in identifying sales leads and poor accuracy of the identified sales leads. Summary of the Invention

[0003] In view of the technical problems of low efficiency in identifying sales leads and poor accuracy of the identified sales leads in the prior art, the present invention provides a method and device for identifying sales leads based on a knowledge graph to solve these problems. The technical solutions of the present invention for solving the above technical problems are as follows: In a first aspect, the present invention provides a method for identifying sales leads based on a knowledge graph, including: Collecting user feature information of a target user and associated user feature information of multiple associated users, where there are multiple associated relationship categories between the target user and the multiple associated users; retrieving in a sales user database constructed based on a knowledge graph based on the multiple associated relationship categories to obtain a set of matching purchased user feature information and multiple sets of matching associated user feature information; performing lead analysis on the user feature information and the multiple associated user feature information according to the set of matching purchased user feature information and the multiple sets of matching associated user feature information to obtain the lead degree of the target user and generate a sales lead identification result, where the lead analysis includes performing similarity analysis on the user feature information and the associated user feature information and performing fusion processing according to the associated relationship category information and the information time.

[0004] Optionally, collecting user feature information of a target user and associated user feature information of multiple associated users includes: collecting user feature information of the target user, where the target user is a user to be identified for sales leads; obtaining multiple associated users of the target user, where there are multiple associated relationship categories between the target user and the multiple associated users; collecting user feature information of the multiple associated users to obtain multiple sets of associated user feature information.

[0005] Optionally, based on multiple association relationship categories, retrieve in the sales user database constructed based on the knowledge graph to obtain a set of feature information of matching purchased users and multiple sets of feature information of matching associated users, including: calling the sales user database constructed based on the knowledge graph, where the sales user database includes feature information of multiple purchased users and associated users; based on multiple association relationship categories, retrieve in the sales user database constructed based on the knowledge graph to obtain multiple purchased users with multiple association relationship categories as multiple matching purchased users, and obtain multiple sets of matching associated users; obtain the user feature information of multiple matching purchased users and multiple sets of matching associated users to obtain a set of feature information of matching purchased users and multiple sets of feature information of matching associated users.

[0006] Among them, the construction steps of the sales user database include: obtaining multiple historical purchased users within a historical time period, and obtaining the historical associated users and association relationship categories of the multiple historical purchased users; collecting the user feature information of the multiple historical purchased users and historical associated users, and combining the corresponding association relationship categories to construct the sales user database, where when the association relationship category or user feature information of the purchased user and the historical associated user changes, the sales user database is updated.

[0007] Optionally, according to the set of feature information of matching purchased users and multiple sets of feature information of matching associated users, perform lead analysis on the user feature information and multiple sets of associated user feature information to obtain the lead degree of the target user and generate a sales lead identification result, including: calculating the similarity between the user feature information and the feature information of each matching purchased user to obtain multiple user similarities, and calculating the mean to obtain the average user similarity; screening the first set of associated user feature information corresponding to the first association relationship category from the multiple sets of associated user feature information, and screening the first set of matching associated user feature information corresponding to the first association relationship category from the multiple sets of matching associated user feature information; calculating the similarity between the first set of associated user feature information and each first set of matching associated user feature information in the first set of matching associated user feature information to obtain a set of first association similarities; continuing to calculate to obtain multiple sets of association similarities for multiple association relationship categories, and calculating the mean to obtain multiple average association similarities; performing weighted fusion calculation on the multiple average association similarities according to the multiple association relationship categories and the recording time of the multiple sets of associated user feature information to obtain a fused association similarity; calculating the lead degree of the target user according to the average user similarity and the fused association similarity to generate a sales lead identification result.

[0008] Among them, according to multiple association relationship categories and the recording times of multiple associated user feature information, weighted fusion calculation is performed on multiple average association similarities to obtain a fused association similarity, including: in the sales user database, obtaining multiple user pairs with a first association relationship category, and calculating the proportion of user pairs in each user pair where both users are purchased users, to obtain a first association clue coefficient; continuing to calculate to obtain multiple association clue coefficients; allocating multiple first association weights according to the multiple association clue coefficients; allocating multiple second association weights according to the recording times of the multiple associated user feature information; calculating multiple association weights according to the multiple first association weights and the multiple second association weights, and performing weighted fusion calculation on the multiple average association similarities to obtain a fused association similarity.

[0009] Among them, allocating multiple second association weights according to the recording times of the multiple associated user feature information includes: obtaining multiple associated recording times of the multiple associated user feature information; calculating the time intervals between the multiple associated recording times and the real-time time to obtain multiple time intervals; allocating and calculating multiple second association weights according to the multiple time intervals, where the size of the time interval is negatively correlated with the size of the second association weight.

[0010] In a second aspect, the present invention provides a sales lead identification device based on a knowledge graph, including: A user information collection module, configured to collect user feature information of a target user and associated user feature information of multiple associated users, where there are multiple association relationship categories between the target user and the multiple associated users; A user information retrieval module, configured to retrieve in a sales user database constructed based on a knowledge graph based on multiple association relationship categories to obtain a set of matching purchased user feature information and multiple sets of matching associated user feature information; A sales lead identification module, configured to perform lead analysis on the user feature information and the multiple associated user feature information according to the set of matching purchased user feature information and the multiple sets of matching associated user feature information to obtain the lead degree of the target user, and generate a sales lead identification result, where the lead analysis includes performing similarity analysis on the user feature information and the associated user feature information, and performing fusion processing according to the association relationship category information and the information time.

[0011] Through the above technical solution, user feature information of a target user and associated user feature information of multiple associated users are collected; then, based on multiple association relationship categories, a search is performed in a sales user database constructed based on a knowledge graph to obtain a set of matching purchased user feature information and multiple sets of matching associated user feature information; by assigning a weighted value according to the similarity between historical user feature data and to-be-identified user feature data, the accuracy of sales lead identification is improved. Then, based on the set of matching purchased user feature information and multiple sets of matching associated user feature information, lead analysis is performed on the user feature information and the multiple sets of associated user feature information to obtain the lead degree of the target user, and a sales lead identification result is generated, realizing the automatic identification of sales leads and greatly improving the identification efficiency and accuracy of sales leads.

[0012] In summary, by implementing the present invention, technical effects of improving the sales lead identification efficiency and the accuracy of identified sales leads can be achieved. Brief Description of the Drawings

[0013] Figure 1 It is a schematic flowchart of a method for identifying sales leads based on a knowledge graph provided by the present invention; Figure 2 It is a schematic structural diagram of a device for identifying sales leads based on a knowledge graph provided by the present invention.

[0014] In the drawings, the components represented by the respective reference numerals are as follows: User information collection module 11, user information retrieval module 12, sales lead identification module 13. Detailed Embodiments

[0015] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present invention.

[0016] In the description of the present invention, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more features. In the description of the present invention, "multiple" means two or more, unless otherwise specifically defined.

[0017] In the description of the present invention, the term "for example" is used to mean "serving as an example, illustration, or explanation". Any embodiment described as "for example" in the present invention is not necessarily construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to implement and use the present invention. In the following description, details are set forth for purposes of explanation. It should be understood that those of ordinary skill in the art can recognize that the present invention can be implemented without these specific details. In other instances, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but rather to be in line with the broadest scope consistent with the principles and features disclosed in the present invention.

[0018] Embodiment 1, as Figure 1 shown, an embodiment of the present invention provides a method for identifying sales leads based on a knowledge graph, including: S100: Collect user feature information of a target user and associated user feature information of multiple associated users, where there are multiple association relationship categories between the target user and the multiple associated users; S200: Based on multiple association relationship categories, retrieve in a sales user database constructed based on a knowledge graph to obtain a set of matching purchased user feature information and multiple sets of matching associated user feature information; S300: According to the set of matching purchased user feature information and multiple sets of matching associated user feature information, perform lead analysis on the user feature information and the multiple associated user feature information to obtain the lead degree of the target user and generate a sales lead identification result, where the lead analysis includes performing similarity analysis on the user feature information and the associated user feature information and performing fusion processing according to the association relationship category information and the information time.

[0019] In an embodiment of the present application, step S100 further includes: Collect user feature information of the target user, where the target user is a user for which sales lead identification is to be performed; Obtain multiple associated users of the target user, where there are multiple association relationship categories between the target user and the multiple associated users; Collect user feature information of the multiple associated users to obtain multiple sets of associated user feature information.

[0020] In the embodiments of the present application, collecting the user feature information of the target user is to obtain multiple associated users associated with the target user according to the feature information of the target user, so as to obtain the information of the associated users and expand the source of sales leads. Among them, the target user can be a user who has ever purchased relevant products, a user who has ever browsed relevant products, or other types of target users. The feature information of the target user is feature information that can to a certain extent reflect the shopping habits of the user and provide a basis for obtaining sales leads, such as age, gender, visit times, consumption records, etc.

[0021] Next, it is necessary to obtain multiple associated users of the target user. An associated user is another user who has certain specific social relationships with the target user. These social relationships can be divided into multiple associated relationship categories according to the relationship category and the degree of relationship intimacy, such as users who have a follow relationship with the target user (which can be divided into unilateral follow and mutual follow), users with different chat durations with the target user (such as chatting continuously for 3 days / 7 days / 30 days, etc.). After obtaining multiple associated users of the target user, it is necessary to collect the feature information of the multiple associated users, and the category of the feature information of the collected associated users is the same as the category of the feature information of the collected target user.

[0022] For the foregoing specific target user and its multiple associated users, as well as the feature information of both, the relationship category between the two, etc., the information can be directly obtained from the data platforms (such as shopping platforms, social platforms, sales settlement systems, etc.) applying the present invention. All this information is voluntarily uploaded by customers and authorized for the platform to use. The specific obtaining method is the prior art and will not be elaborated here.

[0023] In the embodiments of the present application, step S200 further includes: Invoking a sales user database constructed based on a knowledge graph, where the sales user database includes the feature information of multiple purchased users and associated users; Based on the multiple associated relationship categories, retrieving in the sales user database constructed based on the knowledge graph to obtain multiple purchased users having the multiple associated relationship categories as multiple matching purchased users, and obtaining multiple matching associated user sets; Obtaining the user feature information of the multiple matching purchased users and multiple matching associated user sets to obtain a matching purchased user feature information set and multiple matching associated user feature information sets.

[0024] In the embodiments of the present application, constructing a sales user database is to retrieve the data of sales users based on a knowledge graph to realize the subsequent identification of sales leads.

[0025] Among them, the construction steps of the sales user database include: Obtain multiple historical purchased users within a historical time period, and obtain the historical associated users and associated relationship categories of the multiple historical purchased users; Collect the user feature information of multiple historical purchased users and historical associated users, and construct a sales user database in combination with the corresponding associated relationship categories. Wherein, when the associated relationship category or user feature information of the purchased users and historical associated users changes, the sales user database is updated.

[0026] In the embodiment of the present application, the sales user database is built using a knowledge graph. Among them, the categories and acquisition methods of the data included in the sales user database are the same as those in step S100.

[0027] Specifically, the knowledge graph effectively processes, processes, and integrates the data of complex documents, and transforms it into a simple and clear "entity, relationship, entity" triple. Finally, a large amount of knowledge is aggregated to achieve rapid response and reasoning of knowledge. In the knowledge graph, if there is a relationship between two nodes, they will be connected by an undirected edge. Then this node is called an entity, and the edge between them is called a relationship. The basic unit of the knowledge graph is the triple composed of "entity (Entity)-relationship (Relationship)-entity (Entity)", which is also the core of the knowledge graph. Among them, an entity refers to a certain thing with distinguishability and independent existence. For example, in the embodiment of the present application, an entity can be a purchased user and its associated user. An entity is the most basic element in the knowledge graph, and there are different relationships between different entities. A relationship is to connect different entities and refers to the connection between entities. The nodes in the knowledge graph are connected by relationship nodes to form a relationship network. For example, in the embodiment of the present application, the relationship can be social relationship categories such as mutual attention and unilateral attention between entities.

[0028] Therefore, in the sales user database in the embodiments of the present application, data of two types, namely "entities" and "relationships", should also be included. Specifically, the "entities" can be multiple historical purchased users and multiple historical associated users of the historical purchased users. For example, user A who has purchased a certain product within 1 year and associated user B of user A within 1 year. In addition, the historical purchased users and their historical associated users input into the knowledge graph database should also include certain characteristic information, such as information that can reflect their consumption tendencies, such as age, gender, number of visits, consumption records, etc. The "relationship" is the relationship category between the aforementioned historical purchased users and their historical associated users. For example, the relationship category between user A and associated user B of user A can be the type of attention between the two parties (unilateral attention or mutual attention), or the chat duration between the two parties (such as continuous chat for 3 days / 7 days / 30 days), etc. Or, the associated relationship category can also be the interpersonal relationship determined based on user authorization information, such as "spouse", "child", etc.

[0029] Furthermore, in the construction of the sales user database based on the knowledge graph, Neo4j can be used for construction. The specific implementation method can be as follows: Data preparation, the data required here is the aforementioned user data obtained from the data platform (which can be a MySQL table or a CSV file); data cleaning and alignment, that is, unifying the field names from different sources, distinguishing users with the same name, filling in missing fields such as age and gender, etc.; ontology design, defining the User entity class and attributes in Protege, setting relationship types and constraints such as attention type and chat duration (such as relationship time range ≤ 1 year); data import, using Apache NiFi to convert the user attribute table (CSV / MySQL) into node data and the relationship table into edge data, and batch importing them into Neo4j; index optimization, establishing indexes for frequently queried fields (such as number of visits, total consumption amount) to accelerate retrieval. After the construction of the sales user database is completed, relevant users (such as associated users who interact frequently with the target user) can be found through Cypher. In addition, the behavior of the target user (such as new consumption records) can be monitored through Kafka, and the user attributes and relationship weights in Neo4j can be updated in real time. Furthermore, visualization tools such as Gephi or Neo4j Browser can be integrated on the basis of the sales user database to intuitively display the user association network.

[0030] After the sales user database constructed based on the knowledge graph is completed, the feature information of users (such as age, gender, etc.) can be directly called in the database, and sales lead identification can be carried out by comparing the feature information. Specifically, through multiple association relationship categories (such as attention type, chat duration, etc.), retrieval can be performed in the sales user database constructed based on the knowledge graph to obtain multiple purchased users with the multiple association relationship categories (that is, users who have purchased a certain product and are included in the sales user database), as multiple matching purchased users, and obtain multiple matching associated user sets (that is, associated users who have the relationship in the foregoing relationship categories with the foregoing purchased users). For example, when querying the association relationship category as "mutual attention", the feature information set of all purchased users with the association relationship category of "mutual attention" and the corresponding multiple matching associated user feature information sets can be matched from the sales user database, so as to facilitate the subsequent sales lead identification.

[0031] In the embodiment of the present application, step S300 further includes: Calculate the similarity between the user feature information and each matching purchased user feature information to obtain multiple user similarities, and calculate the mean value to obtain the average user similarity; In the multiple associated user feature information, screen the first associated user feature information corresponding to the first association relationship category, and screen the first matching associated user feature information set corresponding to the first association relationship category in the multiple matching associated user feature information sets; Calculate the similarity between the first associated user feature information and each first matching associated user feature information in the first matching associated user feature information set to obtain a first association similarity set; Continue to calculate to obtain multiple association similarity sets of multiple association relationship categories, and calculate the mean value to obtain multiple average association similarities; According to the multiple association relationship categories and the recording time of the multiple associated user feature information, perform weighted fusion calculation on the multiple average association similarities to obtain a fusion association similarity; According to the average user similarity and the fusion association similarity, calculate the lead degree of the target user to generate a sales lead identification result.

[0032] In the embodiments of the present application, calculating the similarity between the user feature information and each piece of the matched purchased user feature information is to calculate the similarity between the feature information of the purchased users and the user feature information (i.e., the user feature information to be predicted currently) among multiple association relationship categories. This similarity can evaluate the credibility of the purchase intention of the user to be predicted identified based on the feature information of the purchased users, so as to improve the accuracy of sales lead identification. Specifically, each feature information needs to be quantified, and then the ratio of each user feature information to each piece of the matched purchased user feature information is calculated, and then the similarity is calculated by weighting. For example, if there are 4 types of feature information (age, gender, visit times, consumption record), a weighting value of 25% can be assigned to each feature. For the similarity between each pair of feature information, the smaller feature value can be divided by the larger feature value and then multiplied by 25%. For age, if the age of the user to be predicted is 25 years old and the age of a purchased user is 50 years old, then the similarity is (25 / 50)*25% = 12.5%. The visit times and consumption record (times) can also be calculated in the same way. For features that are not easy to quantify, direct assignment can be made. For example, if the genders are the same, the similarity is 25%, and if the genders are different, the similarity is 0. Calculate the similarities of the foregoing four types of features respectively, and then accumulate them to obtain the similarity between a certain user feature information and a piece of the matched purchased user feature information. Repeat this process to obtain multiple user similarities, and calculate the average of the multiple user similarities to obtain the average user similarity. Obviously, the average user similarity is any value between 0 and 100%. This value reflects the overall similarity between the user feature information to be identified and the matched purchased user feature information.

[0033] In the embodiments of the present application, it is necessary to screen out the first associated user feature information corresponding to the first association relationship category from multiple associated user feature information, and screen out the first matched associated user feature information set corresponding to the first association relationship category from multiple sets of matched associated user feature information. Specifically, since each user to be identified has multiple types of associated users according to different association relationship categories, there are also corresponding multiple different types of associated user feature information. Therefore, it is necessary to screen out the associated user feature information that belongs to the same relationship category (such as two-way attention) as the user to be identified according to the specific association relationship (i.e., the foregoing first association relationship, such as two-way attention), and screen out the corresponding first matched associated user (i.e., only screen out the matched associated user feature information of two-way attention) feature information set from multiple sets of matched associated user feature information, so as to calculate the foregoing average association similarity.

[0034] In an embodiment of the present application, the similarity between the first associated user feature information and each first matching associated user feature information in the first matching associated user feature information set is calculated to obtain a first associated similarity set, and then multiple associated similarity sets of multiple associated relationship categories are continuously calculated, and the mean value is calculated to obtain multiple average associated similarities, in order to measure the similarity between the associated user feature information corresponding to the user feature information to be recognized and the matching associated user feature information, so as to improve the accuracy of sales lead recognition accordingly. The calculation method of the average associated similarity mentioned here is the same as the calculation method of the aforementioned average user similarity, which will not be elaborated here.

[0035] Among them, according to the multiple associated relationship categories and the recording time of the multiple associated user feature information, a weighted fusion calculation is performed on the multiple average associated similarities to obtain a fused associated similarity, including: In the sales user database, multiple user pairs with the first associated relationship category are obtained, and the proportion of user pairs in which both users in each user pair are purchased users is calculated to obtain a first associated lead coefficient; Multiple associated lead coefficients are continuously calculated; According to the multiple associated lead coefficients, multiple first associated weights are allocated; According to the recording time of the multiple associated user feature information, multiple second associated weights are allocated; According to the multiple first associated weights and multiple second associated weights, multiple associated weights are calculated, and a weighted fusion calculation is performed on the multiple average associated similarities to obtain a fused associated similarity.

[0036] Specifically, obtaining multiple pairs of users in the first associated relationship category is to determine the size of the transaction intention of users under this associated relationship category based on the proportion of both parties in the pairs of users with this associated relationship who have purchased a certain product within the historical time, so as to obtain the first associated clue coefficient and provide a basis for sales lead identification. For example, in a certain first associated relationship category (such as two-way attention), the number of pairs of users where both users in each pair of users included in the sales user database are users who have made purchases is 80, and the total number of pairs of users in this associated relationship category is 100, then the corresponding first associated clue coefficient can be 80 / 100 = 0.8. Through the above method, multiple associated clue coefficients can be calculated. Then, according to the different associated clue coefficients of different associated relationship categories, different weights can be assigned to each associated relationship category. The specific assignment method can be to divide the associated clue coefficient of each associated relationship category by the average value of the associated clue coefficients of all associated relationship categories. For example, the associated clue coefficient of a certain associated relationship category is 0.8, and the average value of the associated clue coefficients of all associated relationship categories is 0.5, then the first associated weight of the associated relationship category is 0.8 / 0.5 = 1.6. From this, multiple first associated weights can be calculated, and the first associated weight can reflect the likelihood of users purchasing a certain product under this associated relationship category (compared with other associated relationship categories).

[0037] Next, it is necessary to assign and obtain multiple second associated weights according to the recording times of multiple associated user feature information records.

[0038] Among them, assigning and obtaining multiple second associated weights according to the recording times of the multiple associated user feature information includes: Obtaining multiple associated recording times of the multiple associated user feature information; Calculating the time intervals between the multiple associated recording times and the real-time time to obtain multiple time intervals; According to the multiple time intervals, calculating and assigning multiple second associated weights, where the size of the time interval is negatively correlated with the size of the second associated weight.

[0039] In the embodiments of the present application, the user information used for sales lead analysis in commercial activities often has timeliness. Generally speaking, the newer the information obtained, the more it can represent the current user consumption psychology and consumption habits, and the higher the reliability. And the older the information, the lower its reliability will gradually be as time goes by. Therefore, in order to ensure the accuracy of sales lead identification, the second associated weight is introduced in this embodiment, and the time interval between the associated user feature information and the real-time time is used to assign weights to the associated user feature information, so as to realize the reasonable invocation of feature information and maximize the accuracy of sales lead identification based on the limited existing information.

[0040] Specifically, when obtaining the characteristic information of associated users, the time of information acquisition (such as a certain year, month, and day) should be recorded and saved together in the aforementioned sales user database. Then, when identifying sales leads, calculate the time interval (calculated in days) between the real-time time and each associated record time to obtain multiple time intervals. Then, calculate multiple second associated weights according to the time interval. The size of the time interval is negatively correlated with the size of the second associated weight, that is, the smaller the time interval, the newer the characteristic information of the associated user, and the higher the second associated weight. For example, if the associated record times of the characteristic information of two associated users are 10 days ago and 15 days ago, the two time intervals are 10 days and 15 days. Then, the second associated weight of the 10-day time interval is the ratio of 1 / 10 divided by the sum of 1 / 10 and 1 / 15, which is 0.6. That is, calculate the ratio of the reciprocal of each time interval to the sum of the reciprocals of multiple time intervals as the second associated weight.

[0041] Optionally, since the data saved in the sales user database is the sales user data within a certain historical time, assume that the data saved in the sales user database is the data within 100 days (that is, the maximum storage is 100 days). Then, the calculation method of the second associated weight can be the remaining storage days of the characteristic information of the associated user divided by the average value of the storage days of all the characteristic information of the associated users in the database. For example, if the characteristic information of a certain associated user has been saved for 10 days and the time interval is 10 days, then the remaining storage days are 100 - 10 = 90 days. If the average value of the storage days of all the characteristic information of the associated users in the database is 50 days, then the corresponding second associated weight is 90 / 50 = 1.8. Through the above method, multiple second associated weights can be calculated and allocated according to multiple time intervals. The smaller the time interval, the larger the second associated weight.

[0042] Next, it is necessary to calculate multiple associated weights based on multiple first associated weights and multiple second associated weights, and perform a weighted fusion calculation on the multiple average association similarities (1 - 100%) to obtain a fusion association similarity.

[0043] Among them, the first associated weight can represent the probability of an associated user purchasing a certain product under a certain association relationship category, and the second associated weight can represent the timeliness of the characteristic information of the associated user, or it can be said to be the credibility.

[0044] The correlation weight obtained by multiplying the corresponding second correlation weight by the first correlation weight can measure the purchase intention of the current user similar to the characteristic information of the associated users entered into the consumer user database at a certain moment under a certain correlation relationship category (such as two-way attention). The larger the value of this correlation weight, the greater the possibility of the current user's purchase, etc. For example, if the first correlation weight is 1.6 and the second correlation weight is 1.8, then the correlation weight is 1.6 * 1.8 = 2.88. Among them, the characteristic information of the associated users under a certain correlation relationship category corresponds to a first correlation weight and a second correlation weight (corresponding to multiple record times of matching the characteristic information of the associated users). In this way, multiple correlation weights of multiple associated users will be calculated and obtained.

[0045] Furthermore, since the average correlation similarity represents the similarity between the associated users and the matching associated users, the higher this similarity, the more reliable the predicted purchase possibility. Therefore, by using multiple correlation weights to perform weighted calculation on multiple average correlation similarities, a fused correlation similarity can be obtained, which can comprehensively consider the probability of the target user purchasing a certain product under the influence of the associated users in multiple correlation relationship categories, the timeliness of the characteristic information of the associated users, and the similarity between the associated users and the historical associated users, so as to obtain a parameter that can more comprehensively measure the possibility of the associated users purchasing a certain commodity. Specifically, assuming that on the basis of the previous example, the average correlation similarities of two associated users are (50% and 40%), and the two correlation weights are (such as 2.88 and 2.4), then the fused correlation similarity is the correlation weight (2.88) * the average correlation similarity (50%) + the correlation weight 2.4 * the average correlation similarity 40% = 2.4. The larger this value, the greater the possibility of the target user purchasing a certain commodity and the higher the credibility.

[0046] Finally, according to the average user similarity and the fused correlation similarity, the lead degree of the target user is calculated to generate a sales lead recognition result. Among them, the average user similarity is used to measure the overall similarity between the target user and the matching users. The specific lead degree calculation method can be the average user similarity multiplied by the fused correlation similarity. Assuming that on the basis of the previous example, the average user similarity (60%) is further introduced, then the lead degree of the target user is the fused correlation similarity (2.4) * the average user similarity (60%) = 1.44. The larger the value of this lead degree, the greater the possibility of the target user and its associated users purchasing a certain commodity under the corresponding correlation relationship category. Among them, the lead value comprehensively considers four parameters: the probability of the target user and its associated users purchasing a certain product under a certain correlation relationship category, the similarity between the associated users and the matching associated users, the timeliness of the characteristic information of the associated users, and the overall similarity between the target user and the matching users, and can well reflect the possibility of the target user and its associated users purchasing a certain commodity.

[0047] The calculation methods of the above clue values can all be automatically calculated using a computer program. The specific implementation method is prior art and will not be elaborated here.

[0048] By calculating the clue values of different target users and their associated users under a certain type of association relationship through the above method, the likelihood of the target user and their associated users purchasing a certain product can be predicted. By changing parameters such as the target user, associated user, and type of association relationship, multiple clue values can be calculated. Based on the magnitudes of the clue values, users with a greater likelihood of purchasing can be identified for promotion, achieving efficient and accurate sales lead identification.

[0049] Embodiment 2, as Figure 2 shown, based on the same inventive concept as the method for identifying sales leads based on a knowledge graph provided in Embodiment 1, the embodiment of the present invention further provides a device for identifying sales leads based on a knowledge graph, including: A user information collection module 11, configured to collect user feature information of a target user and associated user feature information of multiple associated users, where there are multiple types of association relationships between the target user and the multiple associated users; A user information retrieval module 12, configured to retrieve in a sales user database constructed based on a knowledge graph based on multiple types of association relationships to obtain a set of matching purchased user feature information and multiple sets of matching associated user feature information; A sales lead identification module 13, configured to perform clue analysis on the user feature information and the multiple associated user feature information according to the set of matching purchased user feature information and the multiple sets of matching associated user feature information to obtain the clue degree of the target user and generate a sales lead identification result, where the clue analysis includes performing similarity analysis on the user feature information and the associated user feature information and performing fusion processing according to the association relationship category information and the information time.

[0050] Further, the user information collection module 11 includes the following execution steps: Collect user feature information of the target user, where the target user is a user for which sales lead identification is to be performed; Obtain multiple associated users of the target user, where there are multiple types of association relationships between the target user and the multiple associated users; Collect user feature information of the multiple associated users to obtain multiple sets of associated user feature information.

[0051] Further, the user information retrieval module 12 includes the following execution steps: Call a sales user database constructed based on a knowledge graph, where the sales user database includes feature information of multiple purchased users and associated users; Retrieve within the sales user database constructed based on the knowledge graph based on multiple association relationship categories to obtain multiple purchased users with multiple association relationship categories as multiple matching purchased users, and obtain multiple sets of matching associated users; Obtain the user feature information of multiple matching purchased users and multiple sets of matching associated users to obtain a set of matching purchased user feature information and multiple sets of matching associated user feature information.

[0052] Among them, the construction steps of the sales user database include: Obtain multiple historical purchased users within the historical time, and obtain the historical associated users and association relationship categories of the multiple historical purchased users; Collect the user feature information of multiple historical purchased users and historical associated users, and combine the corresponding association relationship categories to construct the sales user database. Among them, when the association relationship category or user feature information of the purchased users and historical associated users changes, update the sales user database.

[0053] Furthermore, the sales lead identification module 13 includes the following execution steps: Calculate the similarity between the user feature information and the feature information of each matching purchased user to obtain multiple user similarities, and calculate the mean value to obtain the average user similarity; Within multiple sets of associated user feature information, screen the first set of associated user feature information corresponding to the first association relationship category, and screen the first set of matching associated user feature information corresponding to the first association relationship category within multiple sets of matching associated user feature information; Calculate the similarity between the first set of associated user feature information and each piece of the first set of matching associated user feature information within the first set of matching associated user feature information to obtain the first set of association similarities; Continue to calculate to obtain multiple sets of association similarities for multiple association relationship categories, and calculate the mean value to obtain multiple average association similarities; According to multiple association relationship categories and the recording time of multiple sets of associated user feature information, perform weighted fusion calculation on multiple average association similarities to obtain the fusion association similarity; According to the average user similarity and the fusion association similarity, calculate the lead degree of the target user to generate the sales lead identification result.

[0054] Among them, performing weighted fusion calculation on multiple average association similarities to obtain the fusion association similarity includes: Within the sales user database, obtain multiple pairs of users with the first association relationship category, and calculate the proportion of pairs of users where both users within each pair of users are purchased users to obtain the first association lead coefficient; Continue to calculate to obtain multiple association lead coefficients; Multiple first association weights are allocated and obtained according to multiple associated clue coefficients; Multiple second association weights are allocated and obtained according to the recording times of multiple associated user feature information; Multiple association weights are calculated according to multiple first association weights and multiple second association weights, and weighted fusion calculation is performed on multiple average association similarities to obtain a fused association similarity.

[0055] Among them, allocating and obtaining multiple second association weights according to the recording times of multiple associated user feature information includes: Obtain multiple associated recording times of multiple associated user feature information; Calculate the time intervals between multiple associated recording times and the real-time, and obtain multiple time intervals; Allocate and calculate multiple second association weights according to multiple time intervals, where the size of the time interval is negatively correlated with the size of the second association weight.

[0056] It should be noted that in the above embodiments, the descriptions of the respective embodiments have their own emphases. For parts not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0057] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0058] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded computers, or other programmable data processing devices to generate a machine, so that the instructions executed by the processors of the computer or other programmable data processing devices generate means for implementing the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0059] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements the functions specified in one or more of the processes Figure 1 and / or boxes Figure 1 specified in one or more of the processes and / or boxes.

[0060] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one or more of the processes Figure 1 and / or boxes Figure 1 specified in one or more of the boxes.

[0061] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made to these embodiments by those skilled in the art once they learn of the basic inventive concept.

[0062] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A method for identifying sales leads based on a knowledge graph, characterized in that, The method includes: Collecting user feature information of a target user and associated user feature information of multiple associated users, where there are multiple associated relationship categories between the target user and the multiple associated users; Based on the multiple associated relationship categories, retrieving in a sales user database constructed based on a knowledge graph to obtain a set of matching purchased user feature information and multiple sets of matching associated user feature information; According to the set of matching purchased user feature information and the multiple sets of matching associated user feature information, performing lead analysis on the user feature information and the multiple associated user feature information to obtain the lead degree of the target user and generate a sales lead identification result, where lead analysis includes performing similarity analysis on the user feature information and the associated user feature information and performing fusion processing according to the associated relationship category information and the information time.

2. The method for identifying sales leads based on a knowledge graph according to claim 1, wherein Collecting user feature information of a target user and associated user feature information of multiple associated users includes: Collecting user feature information of a target user, where the target user is a user to be identified for sales leads; Obtaining multiple associated users of the target user, where there are multiple associated relationship categories between the target user and the multiple associated users; Collecting user feature information of the multiple associated users to obtain multiple sets of associated user feature information.

3. The method for identifying sales leads based on a knowledge graph according to claim 1, wherein Based on the multiple associated relationship categories, retrieving in a sales user database constructed based on a knowledge graph to obtain a set of matching purchased user feature information and multiple sets of matching associated user feature information includes: Invoking a sales user database constructed based on a knowledge graph, where the sales user database includes feature information of multiple purchased users and associated users; Based on the multiple associated relationship categories, retrieving in the sales user database constructed based on a knowledge graph to obtain multiple purchased users with the multiple associated relationship categories as multiple matching purchased users, and obtaining multiple sets of matching associated users; Obtaining the user feature information of the multiple matching purchased users and the multiple sets of matching associated users to obtain a set of matching purchased user feature information and multiple sets of matching associated user feature information.

4. The method for identifying sales leads based on a knowledge graph according to claim 3, wherein The construction steps of the sales user database include: Obtaining multiple historical purchased users within a historical time period, and obtaining historical associated users and associated relationship categories of the multiple historical purchased users; Collecting user feature information of the multiple historical purchased users and the historical associated users, and constructing a sales user database in combination with the corresponding associated relationship categories, where when the associated relationship category or user feature information of the purchased user and the historical associated user changes, the sales user database is updated.

5. The method for identifying sales leads based on a knowledge graph according to claim 1, wherein According to the set of matching purchased user feature information and the multiple sets of matching associated user feature information, performing lead analysis on the user feature information and the multiple associated user feature information to obtain the lead degree of the target user and generate a sales lead identification result includes: Calculating the similarity between the user feature information and the feature information of each matching purchased user to obtain multiple user similarities, and calculating the mean to obtain the average user similarity; Among the multiple associated user feature information, filter the first associated user feature information corresponding to the first associated relationship category, and filter the first matching associated user feature information set corresponding to the first associated relationship category from the multiple matching associated user feature information sets; Calculate the similarity between the first associated user feature information and each first matching associated user feature information in the first matching associated user feature information set to obtain a first associated similarity set; Continue to calculate to obtain multiple associated similarity sets for multiple associated relationship categories, and calculate the mean value to obtain multiple average associated similarities; According to the multiple associated relationship categories and the recording times of the multiple associated user feature information, perform weighted fusion calculation on the multiple average associated similarities to obtain a fused associated similarity; According to the average user similarity and the fused associated similarity, calculate the lead degree of the target user to generate a sales lead recognition result.

6. The method for identifying sales leads based on a knowledge graph according to claim 5, wherein According to the multiple associated relationship categories and the recording times of the multiple associated user feature information, perform weighted fusion calculation on the multiple average associated similarities to obtain a fused associated similarity, including: In the sales user database, obtain multiple user pairs with the first associated relationship category, and calculate the proportion of user pairs in which both users in each user pair are purchased users to obtain a first associated lead coefficient; Continue to calculate to obtain multiple associated lead coefficients; According to the multiple associated lead coefficients, allocate to obtain multiple first associated weights; According to the recording times of the multiple associated user feature information, allocate to obtain multiple second associated weights; According to the multiple first associated weights and multiple second associated weights, calculate to obtain multiple associated weights, and perform weighted fusion calculation on the multiple average associated similarities to obtain a fused associated similarity.

7. The method for identifying sales leads based on a knowledge graph according to claim 6, wherein According to the recording times of the multiple associated user feature information, allocate to obtain multiple second associated weights, including: Obtain multiple associated recording times of the multiple associated user feature information; Calculate the time intervals between the multiple associated recording times and the real-time time to obtain multiple time intervals; According to the multiple time intervals, allocate and calculate multiple second associated weights, where the magnitude of the time interval is negatively correlated with the magnitude of the second associated weight.

8. A sales lead identification device based on a knowledge graph, characterized in that, Including: A user information collection module, configured to collect user feature information of a target user and associated user feature information of multiple associated users, where there are multiple associated relationship categories between the target user and the multiple associated users; A user information retrieval module, configured to retrieve in a sales user database constructed based on a knowledge graph based on the multiple associated relationship categories to obtain a set of matching purchased user feature information and multiple sets of matching associated user feature information; A sales lead identification module, which is used to perform lead analysis on the user feature information and multiple associated user feature information according to the matched purchased user feature information set and multiple matched associated user feature information sets, obtain the lead degree of the target user, and generate a sales lead identification result. Among them, the lead analysis includes performing similarity analysis on the user feature information and the associated user feature information, and performing fusion processing according to the association relationship category information and the information time.

Images