Network group user identification method and device, electronic equipment, storage medium and program product

Abstract

The invention provides a network group user identification method and apparatus, an electronic device, a storage medium and a program product, and relates to the technical field of data analysis, first seed group users are accurately screened through member identification configuration information, and the range of the seed group users is expanded according to a first seed feature by constructing the first seed feature, so that the network group user identification efficiency is improved. Therefore, the coverage rate of the seed group users is improved. According to the invention, the second seed feature is constructed according to the second seed group user, so that the full-quantity network group users of the whole-network user database and the historical user database are obtained, and the accuracy and coverage rate of identifying the group users are improved.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to a method, apparatus, electronic device, storage medium, and program product for identifying network group users. Background Technology

[0002] Currently, with the saturation of the operator's personal business market and the stabilization of the household business market, the enterprise business market has broad growth prospects. For enterprise business, there are two main areas of development: first, promoting and marketing services to government and enterprise entities, such as dedicated enterprise lines; and second, promoting and marketing services to members within government and enterprise entities, such as group ringback tones, group unified payment, and group virtual networks.

[0003] At the level of enterprise group members, operators usually develop business based on the enterprise group members already existing in the Enterprise Customer Service Operation Platform (ESOP) system. However, since there is no management of new / outgoing group members, and the information of group members is entirely entered manually by account managers, there is a lack of effective means of management, resulting in poor usability of group member data in the ESOP system.

[0004] The number of members in government and enterprise groups exceeds 10 million, with each account manager dealing with hundreds of groups and nearly ten thousand group members. Collecting contact lists for these group members is extremely difficult. Data sources, including those provided by client companies and compiled from group business transaction numbers, are disorganized. Account managers sometimes selectively import group members. Due to the large workload involved in data collection, the diverse data sources, and the lack of standardized import practices by account managers, the group member data in the ESOP system is incomplete.

[0005] In the ESOP system, limited manpower leads to delays in recording member turnover, and manual statistics are prone to errors, resulting in untimely manual entry. Static management of group member information and untimely updates to personnel entry and exit have led to discrepancies between the ESOP system's member information and the actual situation over time. Therefore, the group member data in the ESOP system is not updated in a timely manner.

[0006] The existing ESOP system has low coverage of group member identification because the group member data is incomplete and not updated in a timely manner. Summary of the Invention

[0007] This invention provides a method, device, electronic device, storage medium, and program product for identifying network group users, in order to solve the problem of low coverage of group member identification in the prior art and improve the coverage of group member identification.

[0008] In a first aspect, the present invention provides a method for identifying network group users, comprising: filtering first seed group users belonging to the identified group, non-group member users not belonging to the identified group, and users to be judged from a historical user database based on member identification configuration information of the identified group, to obtain an initial identification result of historical users; constructing a first seed feature based on the first seed group users, and filtering first target group users belonging to the identified group from the historical user database based on the first seed feature; filtering second target group users from the historical user database and the entire network user database based on the first seed feature; obtaining second seed group users based on the first target group users and the second target group users; the second target group users are network users belonging to the identified group and network users belonging to other groups; the first seed feature includes basic information, location information, call information, and service information of the first seed group user relative to the identified group; constructing a second seed feature based on the second seed group users, and filtering network group users from the entire network user database and the historical user database based on the second seed feature.

[0009] In one embodiment, filtering users belonging to an identified group from a historical user database based on a first seed feature includes: obtaining a first feature set of all historical users in the historical user database based on the first seed feature; inputting the first feature set into a historical user identification model to obtain the probability of historical users output by the historical user identification model; the historical user identification model is trained based on the features of users in the first seed group, the features of non-group members, and the label of whether or not they are users in the first seed group, based on a first preset logistic regression model; the probability of a historical user represents the probability that the historical user belongs to the identified group; and determining the first target group users based on the probabilities of users in the first seed group, non-group members, users to be judged, and historical users.

[0010] In one embodiment, determining the first target group user based on the probabilities of first seed group users, non-group member users, users to be judged, and historical users includes: when a historical user belongs to the first seed group user and the probability of the historical user is greater than a first probability threshold, the historical user is designated as the first target group user; when a historical user belongs to the users to be judged and the probability of the historical user is greater than a second probability threshold, the historical user is designated as the first target group user; when a historical user belongs to a non-group member user and the probability of the historical user is greater than a third probability threshold, the historical user is designated as the first target group user; wherein, the third probability threshold > the second probability threshold > the first probability threshold.

[0011] In one embodiment, selecting users from a second target group based on a first seed feature in a historical user database and a network-wide user database includes: obtaining a second feature set of all network users in the network-wide user database and the historical user database based on the first seed feature; inputting the second feature set into a network-wide user mining model to obtain the target group to which the network user belongs and the first probability of the network user belonging to the target group, as output by the network-wide user mining model; the network-wide user mining model is trained based on the first seed group user features, the features of non-group members, the label of whether they are users of the first seed group, and the label of the candidate group of the first seed group user, based on a second preset logistic regression model; and determining the second target group users based on the initial identification results, the target group, and the first probability.

[0012] In one embodiment, determining a second target group user based on the initial identification result, the target group, and the first probability includes: when the network user belongs to the entire network user database and the network user's first probability is greater than the fourth probability threshold, the network user is identified as a second target group user; when the network user belongs to the historical user database, the target group is consistent with the initial identification result, and the network user's first probability is greater than the second probability threshold, the network user is identified as a second target group user, and the fourth probability threshold is greater than the second probability threshold.

[0013] In one embodiment, network group users include real network group users, potential network group users, and network group users awaiting reassignment. Network group users are selected from the entire network user database and historical user database based on a second seed feature. This includes: obtaining a third feature set of all network users in the entire network user database and historical user database based on the second seed feature; inputting the third feature set into a network user mining model to obtain the updated target group and the second probability that a network user belongs to the updated target group, as output by the network user mining model; if a network user belongs to the entire network user database and the second probability of the network user is greater than a fifth probability threshold, then the network user is considered a potential network group user of the updated target group; if a network user belongs to the historical user database, the updated target group is consistent with the initial identification result, and the second probability of the network user is greater than the fifth probability threshold, then the network user is considered a real network group user of the updated target group; if a network user belongs to the historical user database, the updated target group is inconsistent with the initial identification result, and the second probability of the network user is greater than the fifth probability threshold, then the network user is considered a network group user awaiting reassignment of the updated target group.

[0014] Secondly, the present invention provides a device for identifying network group users, comprising: a first filtering module, configured to filter out first seed group users belonging to the identified group, non-group member users not belonging to the identified group, and users to be judged from a historical user database based on member identification configuration information of the identified group, to obtain an initial identification result of historical users; a second filtering module, configured to construct a first seed feature based on the first seed group users, and filter out first target group users belonging to the identified group from the historical user database based on the first seed feature; filter out second target group users from the historical user database and the entire network user database based on the first seed feature; and obtain second seed group users based on the first target group users and the second target group users; the second target group users are network users belonging to the identified group and network users belonging to other groups; the first seed feature includes basic information, location information, call information, and service information of the first seed group user relative to the identified group; and a third filtering module, configured to construct a second seed feature based on the second seed group users, and filter out network group users from the entire network user database and the historical user database based on the second seed feature.

[0015] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the above-described methods for identifying network group users.

[0016] Fourthly, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described methods for identifying network group users.

[0017] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements any of the above-described methods for identifying network group users.

[0018] The present invention provides a method, apparatus, electronic device, storage medium, and program product for identifying network group users. By using member identification configuration information, it accurately filters first-seed group users. By constructing first-seed features, it expands the scope of seed group users based on these features to obtain second-seed group users, thereby improving the coverage of seed group users. Furthermore, the present invention constructs second-seed features based on the second-seed group users, enabling the acquisition of the full range of network group users from both the entire network user database and the historical user database, thus improving the accuracy and coverage of group user identification. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is one of the flowcharts illustrating the method for identifying network group users provided by the present invention.

[0021] Figure 2 This is the second flowchart illustrating the method for identifying network group users provided by this invention.

[0022] Figure 3 This is a schematic diagram of the process for obtaining the identification results of historical users provided by the present invention.

[0023] Figure 4 This is a schematic diagram of the process for obtaining the first target group user provided by the present invention.

[0024] Figure 5 This is a schematic diagram of the process for obtaining a second target group user provided by the present invention.

[0025] Figure 6 This is a schematic diagram of the process for obtaining second seed group users provided by the present invention.

[0026] Figure 7 This is a schematic diagram of the process for obtaining the final identification results of network group users from the full network user database and the historical user database provided by the present invention.

[0027] Figure 8 This is a schematic diagram of the structure of the network group user identification device provided by the present invention.

[0028] Figure 9 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0030] The following is combined Figures 1-9 This invention describes a method, apparatus, and electronic device for identifying network group users.

[0031] Figure 1 This is one of the flowcharts illustrating the network group user identification method provided by the present invention, such as... Figure 1 and Figure 2 As shown, the method for identifying network group users includes steps S100 to S300, and the specific steps are as follows.

[0032] S100: Based on the member identification configuration information of the identified groups, filter out the first seed group users belonging to the identified groups, non-group member users not belonging to the identified groups, and users to be judged from the historical user database to obtain the initial identification results of historical users.

[0033] The historical user database includes the Enterprise Customer Service Operation Platform (ESOP) system. Identified groups are those whose group information has been stored in the ESOP system. Identified groups include at least one group. The member identification configuration information and initial identification results for identified groups are shown in Table 1.

[0034] Table 1 Member identification configuration information and initial identification results

[0035] S200: Construct a first seed feature based on the first seed group users; filter out the first target group users belonging to the identified group from the historical user database based on the first seed feature; filter out the second target group users from the historical user database and the entire network user database based on the first seed feature; obtain the second seed group users based on the first target group users and the second target group users.

[0036] The second target group user is a network user belonging to the identified group or to other groups; the first seed feature includes the basic information, location information, call information and business information of the first seed group user relative to the identified group.

[0037] First seed features are constructed based on users in the first seed group. These first seed features include basic information, location information, call information, and business information of the first seed group users relative to the identified groups. For example, basic information includes: whether the historical user's working hours match the average working hours of the identified groups; whether the historical user is under 18 years old; the historical user's age; the historical user's network access duration; whether the historical user has other phone numbers under their identification document; and whether the historical user is associated with the business license of the identified groups. Location information includes: whether the historical user's workplace is consistent with the address of the identified groups; whether the historical user's workplace is far from the address of the identified groups; the distance between the historical user's workplace address and the address of the identified groups; whether the historical user's residence is far from the address of the identified groups; and the distance between the historical user's residence and the address of the identified groups. Call information includes: the number of historical users who made calls within identified groups, the number of calls made by historical users within identified groups, the duration of calls made by historical users within identified groups, the number of historical users who made calls with key figures within identified groups, the number of calls made by historical users with key figures within identified groups, the duration of calls made by historical users with key figures within identified groups, the percentage of historical users who made calls within identified groups, the percentage of historical users who made calls within identified groups, the percentage of historical users who made calls within identified groups, the percentage of historical users who made calls within identified groups, whether historical users had calls with identified groups for two consecutive months, and the total number of groups in the historical user's social circle. Business information includes: whether historical users and identified groups have consistent billing, the historical user's billing expenses, the number of times historical users made billing, whether historical users made billing, whether identified groups made billing, whether historical users and identified groups have consistent cloud computing, the number of cloud computing groups for historical users, whether historical users have cloud computing, whether identified groups have cloud computing, whether historical users and identified groups have consistent virtual networks, and whether historical users have virtual networks.

[0038] Based on the first seed feature, the historical user database is further filtered to obtain the first target group users who belong to the identified groups. Based on the first seed feature, the entire network user database and the historical user database (ESOP system) are filtered to determine the target group with the highest probability of each network user belonging to it, as well as the probability of the network user belonging to that target group. This leads to the acquisition of the second target group users whose belonging groups (including the identified groups and other groups besides the identified groups) can be identified.

[0039] S300: Construct second seed features based on second seed group users, and filter out network group users from the whole network user database and historical user database based on the second seed features.

[0040] The second seed features are constructed based on the users of the second seed group. The second seed features include the basic information, location information, call information, and service information of the users of the second seed group relative to the identified groups and other groups.

[0041] Based on the second seed feature, the system filters the entire network user database and the historical user database, updates the target group with the highest probability of each network user belonging to the target group, and the probability of the network user belonging to the target group, thereby identifying network group users.

[0042] Based on the selected network group users and the primary target group users, the final identification results of network group users in the entire network user database and the historical user database are obtained.

[0043] The network group user identification method provided in this invention accurately filters first seed group users through member identification configuration information. By constructing first seed features, the scope of seed group users is expanded based on these features to obtain second seed group users, thus improving the coverage of seed group users. Furthermore, this invention constructs second seed features based on the second seed group users, enabling the acquisition of the full range of network group users from both the entire network user database and the historical user database, thereby improving the accuracy and coverage of group user identification.

[0044] Based on the above embodiments, the process of filtering users belonging to the identified group from the historical user database based on the first seed feature includes: obtaining a first feature set of all historical users in the historical user database based on the first seed feature; inputting the first feature set into a historical user identification model to obtain the probability of the historical user output by the historical user identification model; the historical user identification model is trained based on the features of the first seed group users, the features of non-group members, and the label of whether or not they are users of the first seed group, based on a first preset logistic regression model; the probability of a historical user represents the probability that the historical user belongs to the identified group; and the first target group users are determined based on the probabilities of the first seed group users, non-group members, users to be judged, and historical users.

[0045] Based on the first seed feature, a first feature set is constructed for all historical users relative to the identified group. The first feature set includes the basic information, location information, call information, and service information of all historical users relative to the identified group in the historical user database (ESOP system).

[0046] Features of users in the first seed group and features of non-group members are extracted from the first feature set and used as training samples for the historical user identification model. The training samples are labeled according to whether they belong to the first seed group, resulting in the first labeled training samples. The first preset logistic regression model is trained using the first labeled training samples until it can accurately identify users in the first seed group and non-group members, thus obtaining the historical user identification model. The historical user identification model calculates the probability that a historical user belongs to an identified group based on its features, obtaining the probability of each historical user. The formula for calculating the probability of a historical user is as follows.

[0047] ; in, The probability of historical users, The features of historical users in the first feature set, These are the classification parameters for the logistic regression model.

[0048] Usually, when hour, The historical user was determined to belong to the identified group. When... hour, The historical user was determined not to belong to the identified group. hour, It was determined that it was impossible to ascertain whether a historical user belonged to an identified group.

[0049] like Figure 4 As shown, the first feature set is input into the historical user identification model to obtain the probability of historical users output by the historical user identification model.

[0050] Based on the probability of historical users and the initial identification results obtained from the member identification configuration information, the first target group users (historical users belonging to the identified group) are obtained.

[0051] Based on the probabilities of users in the first seed group, non-group members, users to be judged, and historical users, the first target group users are determined. Specifically, when a historical user belongs to the first seed group and the probability of the historical user is greater than the first probability threshold, the historical user is designated as the first target group user; when a historical user belongs to the users to be judged and the probability of the historical user is greater than the second probability threshold, the historical user is designated as the first target group user; when a historical user belongs to a non-group member and the probability of the historical user is greater than the third probability threshold, the historical user is designated as the first target group user; wherein, the third probability threshold > the second probability threshold > the first probability threshold.

[0052] like Figure 4As shown, for example, the first probability threshold is 0.45, the second probability threshold is 0.5, and the third probability threshold is 0.9. Based on the probabilities of users in the first seed group, non-group member users, users to be judged, and historical users, the first target group users are determined.

[0053] When a historical user is identified as a first seed group user based on the member identification configuration information, and the probability of the historical user output by the historical user identification model is greater than 0.45, the historical user is taken as the first target group user.

[0054] When a historical user is identified as a user to be judged based on the member identification configuration information, and the probability of the historical user output by the historical user identification model is greater than 0.5, the historical user is taken as the first target group user.

[0055] When a historical user is identified as a non-group member based on the member identification configuration information, and the probability of that historical user is greater than 0.9 according to the historical user identification model, the historical user is designated as the first target group user. This indicates that the member identification configuration information is incorrect, and the historical user identification model is accurate.

[0056] This invention, by constructing a historical user identification model, avoids manual screening of group users, improving the accuracy and efficiency of identifying whether historical users belong to an already identified group, and thus improving the accuracy and efficiency of identifying the first target group users. Through a first probability threshold, a second probability threshold, and a third probability threshold, the probability of historical users output by the historical user identification model is accurately fused with the initial identification result, improving the accuracy of obtaining the first target group users.

[0057] Based on the above embodiments, the process of selecting users from the historical user database and the entire network user database based on the first seed feature includes: obtaining a second feature set of all network users in the entire network user database and the historical user database based on the first seed feature; inputting the second feature set into the entire network user mining model to obtain the target group to which the network user belongs with the highest probability and the first probability that the network user belongs to the target group; the entire network user mining model is trained based on the first seed group user features, the features of non-group members, the label of whether they are users of the first seed group, and the label of the candidate group of the first seed group user; and determining the second target group users based on the initial identification results, the target group, and the first probability.

[0058] Features of users in the first seed group and features of non-group members are extracted from the first feature set and used as training samples for the historical user identification model. The training samples are labeled based on the tags of users in the first seed group (identified groups are the most probable affiliation groups), users not in the first seed group, and candidate groups of users in the first seed group (candidate groups are not the most probable affiliation groups). This results in a second set of labeled training samples. The second pre-set logistic regression model is trained using these labeled training samples until it can accurately output the most probable identified group to which a historical user belongs and the corresponding probability, thus obtaining the overall network user mining model.

[0059] like Figure 5 As shown, a second feature set is constructed for all network users relative to the identified groups and other associated groups based on the first seed features. This second feature set is then input into the network-wide user mining model to obtain the probability of historical users output by the model. The target group to which the network user belongs and the first probability of the network user belonging to the target group are determined. Based on the initial identification results, the target group, and the first probability, users in the second target group are identified.

[0060] Based on the initial identification results, the target group, and the first probability, users in the second target group are determined. Specifically, when a network user belongs to the entire network user database and the network user's first probability is greater than the fourth probability threshold, the network user is identified as a user in the second target group. When a network user belongs to the historical user database, the target group is consistent with the initial identification results, and the network user's first probability is greater than the second probability threshold, the network user is identified as a user in the second target group, and the fourth probability threshold is greater than the second probability threshold.

[0061] like Figure 5 As shown, for example, the second probability threshold is 0.5, and the fourth probability threshold is 0.7. Based on the initial identification results, the target group, and the first probability, the second target group of users is determined.

[0062] When a network user belongs to the entire network user database and the network user's first probability is greater than 0.7, the network user will be regarded as the second target group user.

[0063] When a network user belongs to the historical user database, the target group is consistent with the initial identification result, and the network user's first probability is greater than 0.5, the network user is identified as a second target group user.

[0064] This invention constructs a comprehensive user mining model, avoiding the need for manual screening of target groups with the highest probability of users belonging to specific target groups and the first probability of users belonging to those target groups. This improves the accuracy and efficiency of identifying users in the second target group. By using a second probability threshold and a fourth probability threshold, the target group, the first probability, and the initial identification result output by the comprehensive user mining model are accurately fused, further enhancing the accuracy of acquiring users in the second target group.

[0065] Network group users include real network group users, potential network group users, and network group users awaiting reassignment. Network group users are selected from the entire network user database and historical user database based on a second seed feature. This process includes: obtaining a third feature set of all network users in the entire network user database and historical user database based on the second seed feature; inputting the third feature set into the entire network user mining model to obtain the updated target group and the second probability that the network user belongs to the updated target group; if the network user belongs to the entire network user database and the network user's second probability is greater than the fifth probability threshold, then the network user is considered a potential network group user of the updated target group; if the network user belongs to the historical user database, the updated target group is consistent with the initial identification result, and the network user's second probability is greater than the fifth probability threshold, then the network user is considered a real network group user of the updated target group; if the network user belongs to the historical user database, the updated target group is inconsistent with the initial identification result, and the network user's second probability is greater than the fifth probability threshold, then the network user is considered a network group user awaiting reassignment of the updated target group.

[0066] like Figure 6 As shown, a second seed group user is obtained based on the first target group user and the second target group user.

[0067] The second seed features are constructed based on the users of the second seed group. The second seed features include the basic information, location information, call information, and service information of the users of the second seed group relative to the identified groups and other groups (the target group to which the network user belongs with the highest probability).

[0068] like Figure 7 As shown, a third feature set is constructed based on the second seed features to identify the target group to which all network users have the highest probability of belonging relative to the identified groups. This third feature set is then input into the network-wide user mining model to obtain the updated target group and the second probability of a network user belonging to the updated target group, as output by the model.

[0069] By combining the updated target group, the second probability, and the initial identification results, we can obtain real network group users, potential network group users, and network group users to be assigned.

[0070] like Figure 5 As shown, for example, the fifth probability threshold is 0.35. When a network user belongs to the entire network user database and the network user's second probability > 0.35, then the network user is considered a potential network group user of the updated target group. Further, the potential network group users of the updated target group include first potential network group users, second potential network group users, and third potential network group users. When a network user belongs to the entire network user database and 0.45 > the network user's second probability > 0.35, then the network user is considered a first potential network group user of the updated target group; when 0.6 > the network user's second probability > 0.45, then the network user is considered a second potential network group user of the updated target group; when the network user's second probability > 0.6, then the network user is considered a third potential network group user of the updated target group.

[0071] If a network user belongs to the historical user database (ESOP system), the updated target group is consistent with the initial identification result, and the network user's second probability is greater than 0.35, then the network user is considered as a real network group user of the updated target group.

[0072] If a network user belongs to the historical user database, the updated target group is inconsistent with the initial identification result, and the network user's second probability is greater than 0.35, then the network user will be regarded as a network group user to be assigned to the updated target group.

[0073] Furthermore, the users of the first target group are considered as the real network group users of the identified group. Based on the updated potential network group users of the target group, the updated real network group users of the target group, the updated network group users to be reassigned of the target group, and the real network group users of the identified group, the final identification results of network group users in the entire network user database and the historical user database are obtained.

[0074] The following describes the identification device for network group users provided by the present invention. The identification device for network group users described below can be referred to in correspondence with the identification method for network group users described above.

[0075] like Figure 8 As shown, a device for identifying network group users includes: a first filtering module 801, used to filter out first seed group users belonging to the identified group, non-group member users not belonging to the identified group, and users to be judged from the historical user database according to the member identification configuration information of the identified group, so as to obtain the initial identification results of historical users.

[0076] The second filtering module 802 is used to construct a first seed feature based on the first seed group users, filter out first target group users belonging to the identified group from the historical user database based on the first seed feature, filter out second target group users from the historical user database and the whole network user database based on the first seed feature, and obtain second seed group users based on the first target group users and the second target group users; the second target group users are network users belonging to the identified group and other groups; the first seed feature includes the basic information, location information, call information and service information of the first seed group users relative to the identified group.

[0077] The third filtering module 803 is used to construct second seed features based on the second seed group users, and to filter out network group users from the whole network user database and the historical user database based on the second seed features.

[0078] The network group user identification device provided in this invention accurately filters first seed group users through member identification configuration information. By constructing first seed features, the scope of seed group users is expanded based on these features to obtain second seed group users, thus improving the coverage of seed group users. Furthermore, by constructing second seed features based on the second seed group users, this invention achieves the acquisition of the full range of network group users from both the entire network user database and the historical user database, thereby improving the accuracy and coverage of group user identification.

[0079] In one embodiment, the second screening module 802 is used to: obtain a first feature set of all historical users in the historical user database based on the first seed feature; input the first feature set into the historical user identification model to obtain the probability of the historical user output by the historical user identification model; the historical user identification model is trained based on the features of the first seed group users, the features of non-group members, and the label of whether or not they are the first seed group users, based on the first preset logistic regression model; the probability of the historical user represents the probability that the historical user belongs to the identified group; and determine the first target group users based on the probabilities of the first seed group users, non-group members, users to be judged, and historical users.

[0080] In one embodiment, the second screening module 802 is used to: when a historical user belongs to a first seed group user and the probability of the historical user is greater than a first probability threshold, designate the historical user as a first target group user; when a historical user belongs to a user to be judged and the probability of the historical user is greater than a second probability threshold, designate the historical user as a first target group user; when a historical user belongs to a non-group member user and the probability of the historical user is greater than a third probability threshold, designate the historical user as a first target group user; wherein, the third probability threshold > the second probability threshold > the first probability threshold.

[0081] In one embodiment, the second screening module 802 is used to: obtain a second feature set of all network users in the entire network user database and the historical user database based on the first seed feature; input the second feature set into the entire network user mining model to obtain the target group to which the network user has the highest probability and the first probability that the network user belongs to the target group, as output by the entire network user mining model; the entire network user mining model is trained based on the first seed group user features, the features of non-group members, the label of whether they are users of the first seed group, and the label of the candidate group of the first seed group user, based on the second preset logistic regression model; and determine the second target group users based on the initial identification result, the target group, and the first probability.

[0082] In one embodiment, the second screening module 802 is used to: when a network user belongs to the entire network user database and the network user's first probability is greater than the fourth probability threshold, identify the network user as a second target group user; when a network user belongs to the historical user database, the target group is consistent with the initial identification result, and the network user's first probability is greater than the second probability threshold, identify the network user as a second target group user, and the fourth probability threshold is greater than the second probability threshold.

[0083] In one embodiment, network group users include real network group users, potential network group users, and network group users awaiting relocation. The third screening module 803 is used to: obtain a third feature set of all network users in the entire network user database and the historical user database based on the second seed feature; input the third feature set into the entire network user mining model to obtain the updated target group and the second probability of the network user belonging to the updated target group output by the entire network user mining model; when the network user belongs to the entire network user database and the second probability of the network user is greater than the fifth probability threshold, then the network user is regarded as a potential network group user of the updated target group; when the network user belongs to the historical user database, the updated target group is consistent with the initial identification result, and the second probability of the network user is greater than the fifth probability threshold, then the network user is regarded as a real network group user of the updated target group; when the network user belongs to the historical user database, the updated target group is inconsistent with the initial identification result, and the second probability of the network user is greater than the fifth probability threshold, then the network user is regarded as a network group user awaiting relocation of the updated target group.

[0084] Figure 9 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 9As shown, the electronic device may include: a processor 910, a communications interface 920, a memory 930, and a communications bus 940, wherein the processor 910, the communications interface 920, and the memory 930 communicate with each other through the communications bus 940. The processor 910 can call logical instructions in the memory 930 to execute a method for identifying network group users. This method includes: filtering users belonging to the identified group, non-group members not belonging to the identified group, and users to be judged from a historical user database based on member identification configuration information of the identified group, to obtain initial identification results for historical users; constructing a first seed feature based on the first seed group users, and filtering users belonging to the identified group from the historical user database based on the first seed feature; filtering users belonging to a second target group from the historical user database and the entire network user database based on the first seed feature; obtaining a second seed group user based on the first target group user and the second target group user; the second target group user is a network user belonging to the identified group and another group; the first seed feature includes basic information, location information, call information, and service information of the first seed group user relative to the identified group; constructing a second seed feature based on the second seed group user, and filtering users belonging to the network group from the entire network user database and the historical user database based on the second seed feature.

[0085] Furthermore, the logical instructions in the aforementioned memory 930 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0086] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the network group user identification method provided by the above methods. The method includes: filtering out first seed group users belonging to the identified group, non-group member users not belonging to the identified group, and users to be judged from a historical user database based on the member identification configuration information of the identified group, to obtain the initial identification result of historical users; constructing a first seed feature based on the first seed group users, and filtering out a first target group user belonging to the identified group from the historical user database based on the first seed feature; filtering out a second target group user from the historical user database and the entire network user database based on the first seed feature; obtaining a second seed group user based on the first target group user and the second target group user; the second target group user is a network user belonging to the identified group and belonging to other groups; the first seed feature includes basic information, location information, call information, and service information of the first seed group user relative to the identified group; constructing a second seed feature based on the second seed group user, and filtering out network group users from the entire network user database and the historical user database based on the second seed feature.

[0087] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the network group user identification method provided by the above methods. The method includes: filtering out first seed group users belonging to the identified group, non-group member users not belonging to the identified group, and users to be judged from a historical user database based on the member identification configuration information of the identified group, to obtain the initial identification result of historical users; constructing a first seed feature based on the first seed group users, and filtering out a first target group user belonging to the identified group from the historical user database based on the first seed feature; filtering out a second target group user from the historical user database and the entire network user database based on the first seed feature; obtaining a second seed group user based on the first target group user and the second target group user; the second target group user is a network user belonging to the identified group and belonging to other groups; the first seed feature includes basic information, location information, call information, and service information of the first seed group user relative to the identified group; constructing a second seed feature based on the second seed group user, and filtering out network group users from the entire network user database and the historical user database based on the second seed feature.

[0088] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0089] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method of identifying a network group user, characterized by, The method comprises: screening, according to the member identification configuration information of the identified group, a first seed group user belonging to the identified group, a non-group member user not belonging to the identified group and a to-be-judged user from a historical user database to obtain an initial identification result of the historical user; constructing a first seed feature based on the first seed group user, screening a first target group user belonging to the identified group from the historical user database based on the first seed feature, screening a second target group user from the historical user database and a whole network user database based on the first seed feature, and obtaining a second seed group user based on the first target group user and the second target group user; the second target group user is a network user belonging to the identified group and belonging to other groups; the first seed feature comprises basic information, position information, call information and service information of the first seed group user relative to the identified group; constructing a second seed feature based on the second seed group user, and screening a network group user from the whole network user database and the historical user database based on the second seed feature.

2. The method of identifying a network group user according to claim 1, wherein, The method comprises: obtaining a first feature set of all historical users in the historical user database based on the first seed feature; inputting the first feature set into a historical user identification model to obtain a probability of the historical user output by the historical user identification model; the historical user identification model is obtained based on a first preset logistic regression model, features of the first seed group user, features of the non-group member user and a label indicating whether the user is the first seed group user; the probability of the historical user represents a probability that the historical user belongs to the identified group; determining the first target group user based on the first seed group user, the non-group member user, the to-be-judged user and the probability of the historical user.

3. The method of identifying network group users according to claim 2, wherein, The method comprises: when the historical user belongs to the first seed group user and the probability of the historical user is greater than a first probability threshold, regarding the historical user as the first target group user; when the historical user belongs to the to-be-judged user and the probability of the historical user is greater than a second probability threshold, regarding the historical user as the first target group user; when the historical user belongs to the non-group member user and the probability of the historical user is greater than a third probability threshold, regarding the historical user as the first target group user; wherein the third probability threshold > the second probability threshold > the first probability threshold.

4. The method of claim 1, wherein the step of identifying the network group user is characterized by, The method comprises: obtaining a second feature set of all network users in the full network user database and the historical user database based on the first seed feature; inputting the second feature set into a full network user mining model to obtain a target group to which the network user belongs with the highest probability and a first probability that the network user belongs to the target group output by the full network user mining model; the full network user mining model is obtained based on a second preset logistic regression model, a first seed group user feature, a feature of a non-group member user, a label of whether being a first seed group user, and a label of a candidate group of the first seed group user; determining the second target group user based on the initial identification result, the target group, and the first probability.

5. The method of identifying network group users according to claim 4, wherein, The method for determining the second target group user based on the initial identification result, the target group, and the first probability comprises: when the network user belongs to the full network user database and the first probability of the network user is greater than a fourth probability threshold, regarding the network user as the second target group user; when the network user belongs to the historical user database, the target group is consistent with the initial identification result, and the first probability of the network user is greater than a second probability threshold, regarding the network user as the second target group user, the fourth probability threshold being greater than the second probability threshold.

6. The method of identifying network group users according to claim 4, wherein, The network group user comprises a real network group user, a potential network group user, and a to-be-returned network group user, and the method for screening a network group user based on the second seed feature in the full network user database and the historical user database comprises: obtaining a third feature set of all network users in the full network user database and the historical user database based on the second seed feature; inputting the third feature set into the full network user mining model to obtain an updated target group and a second probability that the network user belongs to the updated target group output by the full network user mining model; when the network user belongs to the full network user database and the second probability of the network user is greater than a fifth probability threshold, regarding the network user as the potential network group user of the updated target group; when the network user belongs to the historical user database, the updated target group is consistent with the initial identification result, and the second probability of the network user is greater than the fifth probability threshold, regarding the network user as the real network group user of the updated target group; when the network user belongs to the historical user database, the updated target group is inconsistent with the initial identification result, and the second probability of the network user is greater than the fifth probability threshold, regarding the network user as the to-be-returned network group user of the updated target group.

7. A device for identifying network group users, characterized in that, The method comprises: The first screening module is configured to screen, from the historical user database, a first seed group user belonging to the identified group, a non-group member user not belonging to the identified group, and a to-be-judged user according to member identification configuration information of the identified group, to obtain an initial identification result of the historical user. The second screening module is configured to construct a first seed feature based on the first seed group user, screen a first target group user belonging to the identified group from the historical user database based on the first seed feature, screen a second target group user from the historical user database and a whole network user database based on the first seed feature, and acquire a second seed group user based on the first target group user and the second target group user. The third screening module is configured to construct a second seed feature based on the second seed group user, and screen a network group user from the whole network user database and the historical user database based on the second seed feature.

8. An electronic device comprising a memory, a processor, and a computer program stored on the memory and running on the processor, characterized in that, The processor executes the computer program to implement the network group user identification method in any one of claims 1 to 6. 9.A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the network group user identification method in any one of claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the network group user identification method in any one of claims 1 to 6.

Images