Key character recognition method and device, equipment, storage medium and product

Abstract

The invention discloses a key character recognition method and device, equipment, a storage medium and a product, and relates to the technical field of computers, and the key character recognition method comprises the steps: obtaining user feature information in response to a key character recognition request; storing the user feature information to a preset social relation graph through a hash function; predicting the social relation graph through a community network algorithm and a machine learning algorithm to obtain a social relation circle; and based on the social relation circle, according to a preset feature discovery engine, determining a target key figure. According to the method, the Hash function is adopted, the user feature information is stored in the social relation graph to comprehensively capture the social relation of the user, the hidden social relation can be effectively found through combination of a community network algorithm and a machine learning algorithm, and the social relation can be found according to the requirements of specific fields or scenes in combination with the preset feature finding engine. The key figures are screened in a customized manner, and the accuracy of key figure recognition is improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and particularly to a method, apparatus, device, storage medium and product for identifying key persons. Background Art

[0002] The booming development of modern information technology has created a huge social relationship network. In the social relationship network, there are key persons with important influence and potential dissemination power, such as high-influence users, opinion leaders, seed users, etc. These key persons can often effectively influence the behaviors and decisions of people around them. Therefore, identifying key persons in the social relationship network has great value in activities such as word-of-mouth marketing, user experience programs, and new business promotion.

[0003] In the related key person identification technology of social relationship networks, key persons are usually identified by constructing social relationships based on the records of social network platforms. However, the activity records on social network platforms cannot accurately reflect the real social relationships of users, resulting in a low accuracy rate of identified key persons. Summary of the Invention

[0004] The main purpose of this application is to provide a method, apparatus, device, storage medium and product for identifying key persons, aiming to solve the technical problem of low accuracy rate of identified key persons.

[0005] To achieve the above purpose, this application proposes a method for identifying key persons, and the method includes:

[0006] Respond to a key person identification request and obtain user feature information;

[0007] Store the user feature information into a preset social relationship graph through a hash function;

[0008] Predict the social relationship graph through a community network algorithm and a machine learning algorithm to obtain a social relationship circle;

[0009] Based on the social relationship circle, determine target key persons according to a preset feature discovery engine.

[0010] In one embodiment, the step of storing the user feature information into a preset social relationship graph through a hash function includes:

[0011] Obtain user account information;

[0012] Establish a mapping relationship set between the user account information and the user feature information;

[0013] Store each element of the mapping relationship set into a preset social relationship graph through a hash function.

[0014] In one embodiment, the step of predicting the social relationship graph through the community network algorithm and the machine learning algorithm to obtain the social relationship circle includes:

[0015] Preliminarily divide the social relationship graph through the community network algorithm, divide user nodes into different communities according to the tightness of their social connections, and obtain a preliminary social relationship circle;

[0016] Based on the preliminary social relationship circle, extract the community feature information of each community;

[0017] Input the community feature information into a preset relationship circle prediction model, predict and optimize the social relationship circle, and obtain the social relationship circle.

[0018] In one embodiment, the feature discovery engine is obtained by fusing multiple models. The step of determining the target key person based on the social relationship circle according to the preset feature discovery engine includes:

[0019] Extract the feature of the user node in the social relationship circle to obtain the user behavior feature;

[0020] Input the user behavior feature into the feature discovery engine, evaluate and rank the user nodes, and obtain the user node ranking result;

[0021] Select the target key person from the user node ranking result according to the preset screening criteria.

[0022] In one embodiment, after the step of selecting the target key person from the user node ranking result according to the preset screening criteria includes:

[0023] Obtain the social interaction data and business feedback data of the key person;

[0024] Analyze and evaluate the social interaction data and the business feedback data to obtain an accuracy evaluation result;

[0025] Adjust the feature discovery engine based on the accuracy evaluation result.

[0026] In one embodiment, after the step of obtaining the user feature information in response to the key person recognition request includes:

[0027] Judge the sensitive data type of the user feature information;

[0028] Select a corresponding desensitization method to desensitize the user feature information according to the sensitive data type.

[0029] In addition, to achieve the above object, the present application also proposes a key person recognition device, which includes: an acquisition module, configured to acquire user feature information in response to a key person recognition request;

[0030] a storage module, configured to store the user feature information into a preset social relationship graph through a hash function;

[0031] a prediction module, configured to predict the social relationship graph through a community network algorithm and a machine learning algorithm to obtain a social relationship circle;

[0032] a determination module, configured to determine a target key person based on the social relationship circle according to a preset feature discovery engine.

[0033] In addition, to achieve the above object, the present application also proposes a key person recognition device, which includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, where the computer program is configured to implement the steps of the key person recognition method as described above.

[0034] In addition, to achieve the above object, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium, and when the computer program is executed by a processor, the steps of the key person recognition method as described above are implemented.

[0035] In addition, to achieve the above object, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the key person recognition method as described above are implemented.

[0036] One or more technical solutions proposed by the present application have at least the following technical effects:

[0037] In related technologies, social relationships are constructed by recording on social network platforms to identify key persons therein. However, the activity records on social network platforms cannot accurately summarize the real social relationships of users, resulting in low accuracy of the identified key persons. In contrast, in the present application, the user feature information is stored into a preset social relationship graph through a hash function, the social relationship graph is predicted to obtain a social relationship circle, and a target key person in the determined social relationship circle is determined according to a preset feature discovery engine. It can be understood that the present application uses a hash function to store user feature information into a social relationship graph to comprehensively capture the social relationships of users. The combination of a community network algorithm and a machine learning algorithm can effectively discover hidden social relationships, and the combination with a preset feature discovery engine can customize the screening of key persons according to the needs of specific fields or scenarios, improving the accuracy of key person recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The drawings herein are incorporated into and constitute a part of this specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application.

[0039] To more clearly illustrate the technical solutions in the embodiments of the present application or in the prior art, the following briefly introduces the drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0040] Figure 1 It is a schematic flowchart provided for the first embodiment of the key person recognition method of the present application;

[0041] Figure 2 It is a schematic flowchart provided for the second embodiment of the key person recognition method of the present application;

[0042] Figure 3 It is a schematic brief flowchart of the key person recognition method provided for the second embodiment of the present application;

[0043] Figure 4 It is a schematic module structure diagram of the key person recognition device for the embodiment of the present application;

[0044] Figure 5 It is a schematic device structure diagram of the hardware operating environment involved in the key person recognition method for the embodiment of the present application.

[0045] The implementation, functional features, and advantages of the present application will be further described with reference to the embodiments and the drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0047] To better understand the technical solutions of the present application, the following will be described in detail in combination with the drawings in the specification and the specific embodiments.

[0048] The main solution of the embodiment of the present application is as follows:

[0049] In response to a key person recognition request, obtain user feature information;

[0050] Through a hash function, store the user feature information into a preset social relationship graph;

[0051] Through a community network algorithm and a machine learning algorithm, predict the social relationship graph to obtain a social relationship circle;

[0052] Based on the social relationship circle, determine the target key person according to a preset feature discovery engine.

[0053] In this embodiment, the present application takes the key person recognition device as the execution subject. For the convenience of description, it is hereinafter referred to as the "device" for specific description.

[0054] Since the prior art usually constructs social relationships by recording on social network platforms and identifies the key persons therein, but the activity records on social network platforms cannot accurately summarize the user's real social relationships, resulting in a low accuracy rate of the identified key persons.

[0055] The present application provides a solution. Through a hash function, the user feature information is stored in a preset social relationship graph, the social relationship graph is predicted to obtain a social relationship circle, and according to a preset feature discovery engine, the target key person in the determined social relationship circle is found. It can be understood that the present application uses a hash function to store the user feature information in a preset social relationship graph. The social relationship graph can capture the user's connection relationships more comprehensively and provide richer user features. By combining the community network algorithm and the machine learning algorithm, the hidden social relationship patterns can be discovered more effectively, and the accuracy of key person recognition can be improved. Through a preset feature discovery engine, key persons can be customized and screened according to the needs of a specific field or scenario, and the accuracy of key person recognition can be improved.

[0056] Based on this, the embodiments of the present application provide a key person recognition method, referring to Figure 1 , Figure 1 is a schematic flowchart of the first embodiment of the key person recognition method of the present application.

[0057] In this embodiment, the key person recognition method includes steps S10 to S40:

[0058] Step S10, in response to a key person recognition request, obtain user feature information;

[0059] It should be noted that the key person recognition request refers to an instruction or signal for requesting the device to identify users with important influence or specific roles in a specific social network or social circle. This request can be initiated by the user actively or automatically triggered by the device according to preset conditions, aiming to find users who play a key role in social interaction, information dissemination, business promotion, etc., for further analysis and application.

[0060] The user characteristic information refers to various attributes and behavioral data related to the user, which can reflect the characteristics of the user's role, influence, behavior pattern, etc. in the social network. The user characteristic information includes, but is not limited to, the user's basic attributes (such as age, gender, occupation, etc.), social behavior data (such as communication records, text message exchanges, APP usage, interactions on social media, etc.), user preferences and interests (such as topics followed, activities participated in, etc.), the user's emotional tendency (such as positive or negative emotions in comments, feedback, etc.), and so on.

[0061] It can be understood that in the fields of social network analysis and business promotion, etc., identifying key figures is an important task. Key figures play roles such as hubs for information dissemination, opinion leaders, and seed users for business promotion in the social network, and have an important impact on the formation, development of the social circle, and the achievement of business goals.

[0062] Therefore, accurately obtaining user characteristic information in this embodiment is of great significance for responding to the key figure recognition request and discovering key figures in the social circle.

[0063] Step S20: Store the user characteristic information into a preset social relationship graph through a hash function;

[0064] It should be noted that the hash function is a mathematical function that can map input data of any length (such as user characteristic information) to an output value of a fixed length (referred to as a hash value or hash code), and has characteristics such as fast calculation, determinism (the same input always produces the same output), and collision resistance (it is difficult to find two different inputs that produce the same output).

[0065] The social relationship graph is a data structure used to represent the social relationships between users, usually presented in the form of a graph, where nodes represent users and edges represent the social connections between users (such as communication, interaction, following, etc.). By storing the user characteristic information into the social relationship graph, it is possible to better understand and analyze the behavior and influence of users in the social network.

[0066] In a feasible implementation manner, step S20 may include:

[0067] Obtain the user account information;

[0068] Establish a mapping relationship set between the user account information and the user characteristic information;

[0069] Through the hash function, store each element of the mapping relationship set into a preset social relationship graph.

[0070] It should be noted that the user account information refers to the account information registered and used by users in social networking platforms or other relevant systems, including but not limited to the user's username, account ID, mobile phone number, email address, etc. User account information is the identifier of a user's identity in the network, which can uniquely determine a user and serve as the key link connecting user characteristic information and the social relationship graph.

[0071] The mapping relationship set refers to the set of association relationships established between user account information and user characteristic information. Each element in the mapping relationship set represents the mapping relationship between a user account information and the corresponding user characteristic information. Through this mapping relationship, the user's account can be associated with the user's characteristic information, providing basic data support for subsequent social relationship analysis and key person identification.

[0072] Exemplarily, after obtaining the user's information from the database, a username table, a phone number name sub-table, and an address city name table are respectively established according to the type of user information, and the fields in the three tables are merged into a dictionary; the dictionary is used to construct a mapping relationship set from user ID to username, and each element in the mapping relationship set is stored in the social relationship graph using a hash function to form a Hash_map object.

[0073] It can be understood that storing user characteristic information in a preset social relationship graph through a hash function can map the user characteristic information to a hash value of a fixed length, making the storage space more compact, reducing data redundancy, and improving storage efficiency. At the same time, the calculation and retrieval speed of the hash value are relatively fast, which is conducive to quickly storing user characteristic information in the social relationship graph to meet the needs of large-scale social network data processing.

[0074] Utilizing the fast retrieval feature of the hash function, the corresponding user node can be quickly found in the social relationship graph according to the hash value of the user characteristic information, improving the efficiency and accuracy of data retrieval.

[0075] Storing user characteristic information in the social relationship graph through a hash function can facilitate the construction and update of the social relationship graph. When the user characteristic information changes or a new user is added, only the hash value needs to be recalculated and the corresponding node in the social relationship graph needs to be updated, without the need to perform a large-scale reconstruction of the entire graph, simplifying the maintenance work of the social relationship graph and improving the flexibility and scalability of the system.

[0076] The social relationship graph stores rich user characteristic information. Combining the efficient processing ability of the hash function, it can support the analysis of complex social relationships, such as community discovery, social circle division, influence propagation path analysis, etc.

[0077] In this embodiment, by comprehensively analyzing user nodes and their characteristic information in the social relationship graph, the social connections and behavior patterns between users can be deeply explored, key figures and potential values in the social network can be discovered, providing strong support for the management and application of the social network.

[0078] Step S30: Predict the social relationship graph through a community network algorithm and a machine learning algorithm to obtain a social relationship circle.

[0079] It should be noted that the community network algorithm refers to an algorithm used to identify and analyze the community structure in a social network. The community network algorithm can divide users into different communities or groups according to the social connections and interaction relationships between users. Users within these communities usually have a relatively high connection density and similarity.

[0080] The machine learning algorithm refers to an algorithm that automatically learns patterns and rules from data and can make predictions and decisions based on the input data.

[0081] The social relationship circle refers to a group or community composed of users with similar characteristics, interests, or social connections in a social network. Users within the social relationship circle usually have a relatively high interaction frequency and close social connections, can form a relatively independent information dissemination and communication environment, and have an important impact on users' behaviors and influence.

[0082] In a feasible implementation manner, step S30 may include:

[0083] Perform a preliminary division of the social relationship graph through a community network algorithm, divide user nodes into different communities according to the closeness of their social connections, and obtain a preliminary social relationship circle.

[0084] Based on the preliminary social relationship circle, extract the community characteristic information of each community.

[0085] Input the community characteristic information into a preset relationship circle prediction model, predict and optimize the social relationship circle, and obtain the social relationship circle.

[0086] It should be noted that the user node is a node in the social relationship graph, representing an individual user in the social network. User nodes are connected to other nodes through edges, indicating the social connections and interaction relationships between users.

[0087] The closeness of the social connection refers to the intensity and frequency of the social connection between users, which can usually be measured by factors such as the interaction frequency, interaction intensity, and activities participated in together between users. The closeness of the social connection reflects the closeness of the relationship between users and is an important basis for community division.

[0088] The preliminary social relationship circle is a social relationship circle obtained by preliminarily dividing the social relationship graph through a community network algorithm. The preliminary social relationship circle is a community structure preliminarily divided according to the closeness of social connections between user nodes, providing a basis for further prediction and optimization of the social relationship circle.

[0089] The community feature information refers to the information describing the characteristics and attributes of the community, including but not limited to the number of user nodes in the community, the number of edges in the community, the average interaction frequency between users in the community, the common interests and topics of users in the community, the tightness and stability of the community, etc.

[0090] The relationship circle prediction model is a model used to predict and optimize the social relationship circle. The relationship circle prediction model is constructed based on machine learning algorithms. By analyzing community feature information and social relationship data, a prediction model is established to predict and optimize the development and changes of the social relationship circle, so as to obtain a more accurate and stable social relationship circle.

[0091] Exemplarily, according to the social relationship graph, an adjacency matrix for representing the relationship between users and a correlation matrix for measuring the degree of association between users are obtained. Based on social network algorithms such as community discovery and combined with machine learning algorithms, personal social circle discovery is realized.

[0092] The specific algorithm for calculating the correlation matrix is:

[0093]

[0094] where A ij is the matrix representation of the original social graph; is the probability that two nodes i and j may be connected; δ(c i , c j ) represents whether two nodes i and j in the original graph are connected. If they are connected, it is 1, otherwise it is 0.

[0095] The algorithm for realizing personal social circle discovery based on social network algorithms such as community discovery and combined with machine learning algorithms includes two stages, referring to Figure 2 :

[0096] Stage 1: Modularity Optimization. By continuously traversing the nodes in the network of the social relationship graph, try to add a single node to the community that can maximize the modularity improvement until all nodes no longer change.

[0097] Stage 2: Community Aggregation. It will process the results of the first stage, merge small communities into a super node to reconstruct the network. At this time, the edge weight is the sum of the edge weights of all original nodes within the two nodes.

[0098] In the first round of iteration of the algorithm, each node is initially regarded as a separate community. Then, the algorithm optimizes modularity by moving nodes to adjacent communities if such a move can increase modularity. After the first round, the algorithm aggregates each community into a single node, forming a new network. Then, the algorithm repeats the process of the first round on this new network. This process can be repeated multiple times until the modularity no longer increases significantly or a preset number of iterations is reached.

[0099] The algorithm can quickly identify community structures in large-scale networks and, through multiple rounds of iteration, gradually improve the quality of community partitioning.

[0100] Exemplarily, the device preprocesses the social network graph using a random walk algorithm. The random walk algorithm can capture potential connections and social structures between users by simulating the random browsing process of users in the social network, providing more accurate basic data for subsequent node importance ranking and community partitioning. Then, a generalized PageRank algorithm and an algorithm based on a graph convolutional neural network (GNN) are used to rank the importance of nodes. The generalized PageRank algorithm evaluates the influence and importance of nodes in the network by calculating the PageRank values of nodes; while the graph convolutional neural network (GNN) can directly learn graph-structured data using a deep learning framework, fully understand the internal laws and deeper semantic features of nodes and edges, and thus more accurately evaluate the importance of nodes.

[0101] For the i-th node to be measured, take the set of all other nodes in its adjacency graph that contain this node, and select the set of important nodes Bi from it. If Bi is an empty set, end this loop; otherwise, take any member of Bi as the corresponding "test point" node, and repeat the above steps until the requirements are met. Through this method, the potential community structure in the social relationship graph can be initially identified, and user nodes can be divided into different communities according to the closeness of their social connections, obtaining a preliminary social relationship circle.

[0102] Taking all of them as a whole, perform an aggregation operation on the set of all other unexamined nodes in a graph convolutional manner to obtain an aggregated result set, where each element represents a list of important potential partners. Then, traverse all the nodes in the entire graph set and add the corresponding nodes to it as well. Finally, evaluate the similarity degree of these important potential collaborators (i.e., the homogeneity between the candidate object and the current target user) through metrics such as the Levenshtein distance (edit distance) or Jaccard similarity coefficient, further optimizing the partitioning result of the preliminary social relationship circle.

[0103] Based on the preliminary social relationship circle, extract the community feature information of each community. The community feature information includes the inherent attributes of nodes (such as the labels, categories, degrees, etc. of nodes), context information (the positions of nodes in the graph, such as centrality, community structure, etc.), and neighbor information of nodes (the features of the neighbor nodes of the node, which can be aggregated through the message passing mechanism of the GNN).

[0104] Input the extracted community feature information into a preset relationship circle prediction model to predict and optimize the social relationship circle. The relationship circle prediction model can achieve efficient feature extraction based on graph convolution, update the feature representation of nodes by aggregating the information of neighbor nodes, and use different aggregation functions (such as summation, averaging, maximum value, etc.) to achieve feature aggregation. This application adopts a spatial domain-based method, directly utilizes the topological structure of the graph, collects information according to the neighbor information of the graph, has higher computational efficiency, is suitable for processing large-scale graph data, and thus obtains a more accurate and stable social relationship circle.

[0105] It can be understood that in social network analysis, accurately identifying and predicting social relationship circles is of great significance for understanding the structure and function of social networks, and for social network management and operation. This application preliminarily divides the social relationship graph through a community network algorithm, divides user nodes into different communities according to the tightness of their social connections, and obtains a preliminary social relationship circle.

[0106] Next, based on the preliminary social relationship circle, extract the community feature information of each community. The extraction of community feature information is a key step in the prediction and optimization of social relationship circles because these feature information can comprehensively reflect the characteristics and attributes of communities and provide rich input data for the prediction model.

[0107] Finally, input the extracted community feature information into a preset relationship circle prediction model to predict and optimize the social relationship circle, and obtain the final social relationship circle. The relationship circle prediction model can be constructed using various machine learning algorithms, such as decision trees, random forests, neural networks, etc. By training the model to learn the mapping relationship between community features and social relationship circles, predict the development trend and changes of social relationship circles, and optimize the preliminary social relationship circle to improve the accuracy and stability of social relationship circles.

[0108] Step S40, based on the social relationship circle, determine the target key person according to a preset feature discovery engine.

[0109] It should be noted that the feature discovery engine is a system or algorithm for identifying and extracting user features, which can mine information reflecting key features such as user importance, influence, and activity from multi-source information such as user behavior data and social relationship data.

[0110] The target key person is a user with important influence, status or value in a specific social relationship circle or social network. They may be opinion leaders, hubs of information dissemination, key nodes for business promotion, etc. Their behaviors and opinions have a significant guiding and promoting effect on the development of the social relationship circle and the operation of the social network.

[0111] It can be understood that the social relationship circle provides a clear analysis scope and background for the feature discovery engine, enabling the feature discovery engine to more accurately identify the key figures in this social relationship circle. By analyzing the user characteristics and interaction relationships within the social relationship circle, the feature discovery engine can accurately discover those users with high influence, activity and importance within the circle, so as to determine the target key persons.

[0112] The feature discovery engine can evaluate the characteristics of users from multiple dimensions, including the strength of social connections, interaction frequency, information dissemination path, influence dissemination range, etc. By comprehensively analyzing these characteristics, the role and function of users in the social relationship circle can be comprehensively understood, so as to more accurately determine the target key persons. For example, a user may have a large number of fans and followers in the social relationship circle, but his interaction frequency is low and the influence dissemination range is limited. Therefore, he cannot be simply regarded as a target key person; while the feature discovery engine can comprehensively consider these factors and give a more accurate evaluation result.

[0113] Determining the target key persons is of great significance for the management and operation of the social network. The social network platform can formulate targeted content push strategies, activity planning programs and user service plans according to the characteristics and needs of the target key persons, improve user satisfaction and activity, and enhance the attractiveness and competitiveness of the social network. For example, for the target key persons, the social network platform can provide exclusive content customization services, opportunities to participate in activities first, etc. to attract and retain these important users.

[0114] In the business field, understanding and determining the target key persons plays an important role in business promotion and marketing. Enterprises can formulate effective marketing strategies according to the characteristics and influence of the target key persons, such as promoting products and brand endorsement through cooperation with the target key persons, expanding the coverage and influence of the business, and improving the marketing effect and return on investment. For example, a newly launched fashion brand can look for target key persons in the fashion social relationship circle for cooperation, and rely on their influence and word-of-mouth effect to quickly enhance the brand's popularity and reputation.

[0115] In a feasible implementation manner, step S40 may include:

[0116] Extract the characteristics of the user nodes in the social relationship circle to obtain user behavior characteristics;

[0117] Input the user behavior characteristics into the feature discovery engine, evaluate and rank the user nodes, and obtain the user node ranking result;

[0118] Select the target key figures from the user node ranking result according to the preset screening criteria.

[0119] It should be noted that the user node ranking result refers to the influence ranking of user nodes obtained through evaluation and ranking. The user node ranking result shows the relative importance and influence of users in the social relationship circle.

[0120] The screening criteria refer to the preset conditions and standards for selecting target key figures from the user node ranking result. The screening criteria can include influence thresholds, user activity ranges, user behavior characteristic requirements, etc., to ensure that the selected target key figures meet specific business needs and application scenarios.

[0121] Exemplarily, obtain the user behavior data in the social relationship circle from the social network platform, including the user's posting records, comment content, liked objects, forwarding behaviors, private message communication records, etc.

[0122] Clean and format the collected user behavior data, remove invalid and redundant data, and extract valuable information, such as the posting time of the post, the sentiment tendency of the comment, the frequency and object of the like, the range and speed of the forward, etc.

[0123] Extract user behavior characteristics from the preprocessed data, specifically including:

[0124] Count the number of posts, comments, likes, forwards, etc. of the user within a certain period of time, and calculate the user activity index.

[0125] Analyze the number and speed of the content posted by the user being liked, commented, and forwarded by other users, and calculate the user influence score; at the same time, consider the quantity and quality of the user's fans, as well as the activity and influence of the fans, and comprehensively evaluate the influence of the user.

[0126] Construct a social relationship graph between users, represented in the form of a directed weighted graph, where nodes represent users, edges represent relationships such as attention and interaction between users, and the weights on the edges represent the relationship strength. For example, the higher the attention of user u to user v, that is, the greater the relationship strength value, the greater the weight of edge uv.

[0127] Use natural language processing technology to perform sentiment analysis on the text data such as the user's comments and posting content, judge whether the user's sentiment tendency is positive, negative or neutral, and extract the user's sentiment tendency characteristics.

[0128] Input the extracted user behavior features into the feature discovery engine, which is composed of a multi-channel joint feature discovery model, including PageRank model, influence negativity algorithm, HITS algorithm, rule model, etc.

[0129] Use the PageRank model to calculate the PageRank value of each user node, that is, the importance and influence ranking of the user in the social network. The higher the PageRank value, the more important the user is.

[0130] Evaluate the negative influence of users in the social network through the influence negativity algorithm, that is, the degree of influence caused by users posting negative information or spreading negative information, and identify the key figures with potential negative influence.

[0131] Apply the HITS algorithm to calculate the authority value and hub value of each user node. The authority value represents the degree to which the user is concerned by other important users, and the hub value represents the user's ability to concern other important users. Evaluate the user node by combining these two indicators.

[0132] According to the preset rule model, combined with the user's behavior characteristics and social relationship characteristics, evaluate the user node. For example, set the rule "If the number of posts of a user exceeds 10 within a week, and the number of likes per post exceeds 100 on average, then the user is considered to have high activity and influence", and score and rank the user node according to the rule.

[0133] Integrate the evaluation results of the above models and algorithms, consider the user's sentiment tendency characteristics, comprehensively evaluate and rank the user nodes, obtain the user node ranking result, and display the relative importance and influence of the user in the social relationship circle.

[0134] According to the business requirements and application scenarios, preset the screening criteria for target key figures, such as influence threshold (for example, users with the top 10% PageRank value), activity requirements (such as the number of posts within a week is not less than 5), sentiment tendency requirements (such as the proportion of positive sentiment tendency exceeds 80%), etc.

[0135] From the user node ranking result, initially select the users who meet the conditions as candidate target key figures according to the preset screening criteria.

[0136] Further extract and verify the candidate target key figures, select the first unconfirmed person, and include it in the currently extracted sample set.

[0137] Then find the next unconfirmed sample point, and then connect these two nodes to form a new path.

[0138] Repeat the above operations until all the undetermined nodes in the figure are covered.

[0139] First, define an array with a length of L to store the information related to these found candidates.

[0140] Then, set another special character array with a length of S and an initial value of 0 as the identifier array to indicate which candidates have been excluded.

[0141] Then, traverse each element of the special character array and modify the data stored in the array according to the object position it points to.

[0142] Finally, check the performance of each candidate in each dimension in turn to determine the best one among all the optional candidates, or if there is no best choice, select the one closest to the best result.

[0143] After the above key person extraction steps, if the selected samples still do not meet the set target conditions within this cycle, other people need to be reselected for experimental research; usually, some measures can be taken for remedy, such as increasing the number of samples or reducing the test difficulty, etc., to improve the final effect, so as to finally determine the target key person.

[0144] In a feasible implementation manner, after the step of selecting the target key person from the user node sorting result according to the preset screening criteria, the following steps are included:

[0145] Obtain the social interaction data and business feedback data of the key person;

[0146] Analyze and evaluate the social interaction data and the business feedback data to obtain an accuracy evaluation result;

[0147] Based on the accuracy evaluation result, adjust the feature discovery engine.

[0148] It should be noted that the social interaction data refers to the data generated when a user interacts with other users in a social network. The social interaction data includes, but is not limited to, the user's communication records (such as phone call records, text message exchanges), interaction behaviors on social platforms (such as likes, comments, forwards, private message exchanges), records of participated social activities, etc. These data can reflect the social behavior patterns, interaction frequencies, influence propagation paths, emotional capabilities, etc. of users in the social network.

[0149] The business feedback data refers to the feedback information provided by users when participating in business activities or using business products and services. The business feedback data includes, but is not limited to, user feedback on the participation in marketing activities (such as the willingness to participate in activities, the willingness to purchase, the willingness to recommend, etc.), product usage feedback (such as product satisfaction evaluation, functional requirement suggestions, complaint information, etc.), and service experience feedback (such as service quality evaluation, service improvement suggestions, etc.). These data can reflect users' attitudes and needs towards business activities, products, and services, providing a basis for business decision-making and optimization.

[0150] The accuracy evaluation result refers to the result of the accuracy of key person identification obtained through analysis and evaluation. The accuracy evaluation result can reflect the accuracy of the feature discovery engine in identifying key persons, providing a basis for subsequent model optimization and adjustment. For example, the accuracy evaluation result may indicate that the feature discovery engine has achieved a relatively high accuracy rate in identifying key persons with high social influence and high business value.

[0151] It can be understood that the sentiment analysis model is trained using the labeled sentiment dataset, the model parameters are optimized, and the accuracy of sentiment analysis is improved, enabling it to more accurately identify the sentiment tendencies and emotional changes of key persons in social interactions. Combining social interaction data and business feedback data, the multi-model fusion algorithm in the feature discovery engine is adjusted and optimized, such as adjusting the weights of sentiment tendency features in the PageRank algorithm and the GNN algorithm, introducing an attention mechanism to determine the importance of sentiment tendency in influence ranking, and improving the accuracy of the model in identifying key persons.

[0152] The adjusted feature discovery engine is verified through the cross-validation method. The dataset is divided into a training set and a test set, and the model is trained and tested multiple times to evaluate the stability and generalization ability of the model. An A / B test is conducted, and the feature discovery engine before and after adjustment is applied to the actual business scenario to compare the differences in aspects such as the accuracy of key person identification and the business promotion effect between the two, further verifying the performance of the adjusted model.

[0153] Based on the accuracy evaluation result and business feedback, the feature discovery engine is continuously iterated and optimized, and the model parameters and algorithm logic are continuously updated to adapt to the changes in the social network and business environment, improving the accuracy and business value of key person identification.

[0154] This embodiment provides a method for identifying key persons. By combining the community network algorithm and the machine learning algorithm, hidden social relationship patterns can be discovered more effectively, and the accuracy of key person identification can be improved. Through the preset feature discovery engine, key persons can be customized and screened according to the needs of specific fields or scenarios, improving the accuracy rate of key person identification.

[0155] In a feasible implementation, after step S10, the following steps are included:

[0156] Determine the sensitive data type of the user feature information;

[0157] According to the sensitive data type, select the corresponding desensitization method to desensitize the user feature information.

[0158] It should be noted that the sensitive data type refers to the data types in the user feature information that may involve user privacy or security. Sensitive data types usually include personal identity information (such as name, ID number), contact information (such as mobile phone number, email address), financial information (such as bank account, credit card information), location information (such as home address, real-time location), biometric information (such as fingerprint, facial recognition feature), etc. The desensitization method refers to the method of processing sensitive data so that it cannot be recognized or associated with a specific individual.

[0159] The desensitization process refers to the process of processing the sensitive data in the user feature information using the corresponding desensitization method according to the sensitive data type. The purpose of desensitization is to ensure that the user feature information can be safely stored, transmitted, and used while protecting user privacy and data security, and at the same time meeting the requirements of business analysis and applications.

[0160] Exemplarily, the structured data involved in this application can be directly processed and desensitized through database query languages. For example, use SQL statements to replace or encrypt sensitive information in the database.

[0161] Replacement method: Replace sensitive information with other non-sensitive information or placeholders. For example, replace the ID number with a fixed mask form, and for the mobile phone number, display the first three digits and the last four digits of the mobile phone number, and replace the middle digits with asterisks or other characters.

[0162] Randomization method: Generate data with the same format as the original data but with random content to replace sensitive information. For example, randomly generate an email address that conforms to the format.

[0163] Generalization method: Generalize detailed data to a broader category or range. For example, replace the specific age with an age range (such as "20 - 25 years old"). For call records, in addition to desensitizing the mobile phone number, also perform fuzzy processing on the call time, and only display the call time period instead of the specific time.

[0164] Deletion method: Directly delete sensitive fields or records, which is applicable to data that is not important for business analysis.

[0165] For the unstructured data involved in this application, such as text, images, audio, and video, specific technologies and tools are required for processing. For example, natural language processing (NLP) technology is used for the analysis and extraction of text data.

[0166] Using NLP technology, valuable information can be extracted from users' communication records, such as frequently contacted numbers, most used APPs, etc., so as to better understand users' needs and behavior patterns.

[0167] User feedback received by telecom operators often exists in text form, and these feedbacks may cover different business categories. Using NLP technology, especially text classification methods, complaint texts can be automatically assigned to the corresponding business categories. The user feedback text classification application based on the GRW and FastText models in this application constructs an efficient text classification method by extracting effective feature words. Experimental results show that this method can extract structured information from unstructured text data. For example, named entity recognition (NER) can identify entities such as personal names, locations, and organizations in the text. For instance, specific product or service names can be extracted from users' feedback to improve service quality or product functions, and it is superior to other methods in terms of accuracy, Kappa coefficient, and Hamming loss.

[0168] Through NLP technology, sentiment analysis can be performed on users' comments or feedback to determine whether the user's sentiment tendency is positive or negative. The specific steps include:

[0169] Data preprocessing: cleaning the data (removing duplicate comments, advertisements, irrelevant information, etc.), Chinese word segmentation (splitting sentences into words or phrases), removing stop words (removing common meaningless words, such as "de", "le", etc.), and part-of-speech tagging (tagging the part of speech of each word, such as noun, verb, etc.).

[0170] Convert the preprocessed text into numerical features that can be understood by machine learning models. Commonly used feature extraction methods use TF-IDF (term frequency–inverse document frequency), etc.

[0171] Use machine learning or deep learning models for sentiment classification. The models used in this application include Bayesian methods. The model is trained with training set data and then the performance of the model is evaluated using the test set.

[0172] The performance of the model can be evaluated by indicators such as accuracy, recall rate, F1 score, etc. Analyze the sentiment distribution to identify keywords and themes of highly satisfied and lowly satisfied comments.

[0173] Monitor the emotional trend, promptly identify problems in products or services, and take corresponding measures.

[0174] Exemplarily, to facilitate understanding of the implementation process of the key person recognition method obtained by combining the above-mentioned Embodiment 1, please refer to Figure 3 , Figure 3 A brief process schematic diagram of a key person recognition method is provided. Specifically:

[0175] The device first responds to a key person recognition request and obtains user feature information through multiple channels. This information includes the user's personal profile, social interaction records, business feedback, etc. The acquisition of user feature information is to comprehensively understand the user's behavior and influence in the social network.

[0176] After obtaining the user feature information, the device determines the sensitive data type of the user feature information, such as personal identity information, contact information, etc. According to the sensitive data type, a corresponding desensitization method is selected for desensitization processing to protect user privacy and data security.

[0177] The device stores the desensitized user feature information into a preset social relationship graph through a hash function.

[0178] The device uses community network algorithms and machine learning algorithms to predict the social relationship graph to obtain a social relationship circle.

[0179] Feature extraction is performed on the user nodes in the social relationship circle to obtain user behavior features;

[0180] The user behavior features are input into a feature discovery engine to evaluate and rank the user nodes, obtaining a user node ranking result;

[0181] According to the preset screening criteria, target key persons are selected from the user node ranking result.

[0182] The device obtains the social interaction data and business feedback data of the selected key persons, including communication records, text messages, APP usage data, browsing history, marketing activity feedback, product usage feedback, etc. These data are analyzed and evaluated to obtain an accuracy evaluation result.

[0183] Based on the accuracy evaluation result, the device adjusts the feature discovery engine to improve the accuracy of key person recognition. The adjustment may involve modifying model parameters, improving algorithm logic, introducing new feature extraction methods, etc.

[0184] This embodiment can be applied to different social scenarios such as family circles, work social circles, and close friend / hometown circles. In a family circle, the key decision-makers in the family can be identified; in a work social circle, the key contacts can be identified; in a close friend / hometown circle, the key opinion leaders can be identified. In this way, social network analysis and business promotion can be carried out more accurately.

[0185] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the method for identifying key persons in this application. Based on this technical concept, more forms of simple transformations are within the protection scope of this application.

[0186] This application also provides a key person identification device. Please refer to Figure 4 , the key person identification device includes:

[0187] An acquisition module 10, configured to acquire user feature information in response to a key person identification request;

[0188] A storage module 20, configured to store the user feature information into a preset social relationship graph through a hash function;

[0189] A prediction module 30, configured to predict the social relationship graph through a community network algorithm and a machine learning algorithm to obtain a social relationship circle;

[0190] A determination module 40, configured to determine a target key person based on the social relationship circle according to a preset feature discovery engine.

[0191] Optionally, the storage module 20 includes;

[0192] A first acquisition module, configured to acquire user account information;

[0193] A first establishment module, configured to establish a mapping relationship set between the user account information and the user feature information;

[0194] A first storage module, configured to store each element of the mapping relationship set into a preset social relationship graph through a hash function.

[0195] Optionally, the prediction module 30 includes;

[0196] A first partitioning module, configured to perform a preliminary partitioning on the social relationship graph through a community network algorithm, divide user nodes into different communities according to the tightness of their social connections, and obtain a preliminary social relationship circle;

[0197] A first extraction module, configured to extract community feature information of each community based on the preliminary social relationship circle;

[0198] The first prediction module is configured to input the community feature information into a preset relationship circle prediction model, predict and optimize a social relationship circle, and obtain the social relationship circle.

[0199] Optionally, the determination module 40 includes;

[0200] The second extraction module is configured to extract features of user nodes in the social relationship circle to obtain user behavior features;

[0201] The first input module is configured to input the user behavior features into a feature discovery engine, evaluate and rank the user nodes, and obtain a user node ranking result;

[0202] The first selection module is configured to select target key figures from the user node ranking result according to a preset screening criterion.

[0203] Optionally, the determination module 40 includes;

[0204] The second acquisition module is configured to acquire social interaction data and business feedback data of the key figures;

[0205] The first evaluation module is configured to analyze and evaluate the social interaction data and the business feedback data to obtain an accuracy evaluation result;

[0206] The first adjustment module is configured to adjust the feature discovery engine based on the accuracy evaluation result.

[0207] Optionally, the key figure recognition device includes;

[0208] The first judgment module is configured to judge the sensitive data type of the user feature information;

[0209] The first desensitization module is configured to select a corresponding desensitization method to desensitize the user feature information according to the sensitive data type.

[0210] The key figure recognition device provided in this application adopts the key figure recognition method in the above embodiment, and can solve the technical problem of low accuracy of the recognized key figures. Compared with the prior art, the beneficial effects of the key figure recognition device provided in this application are the same as those of the key figure recognition method provided in the above embodiment, and other technical features in the key figure recognition device are the same as those disclosed in the method of the above embodiment, and will not be elaborated here.

[0211] The present application provides a key person recognition device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the key person recognition method in Embodiment 1 above.

[0212] Reference is made below Figure 5 , which shows a schematic structural diagram of a key person recognition device suitable for implementing the embodiments of the present application. The key person recognition device in the embodiments of the present application may include, but is not limited to, mobile terminals such as mobile phones, tablet computers, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PMPs (Portable Media Players), in-vehicle terminals (such as in-vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5 The key person recognition device shown is only an example and should not impose any limitation on the functions and usage scope of the embodiments of the present application.

[0213] As Figure 5 shown, the key person recognition device may include a processing device 1001 (such as a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM: Read Only Memory) 1002 or the program loaded from the storage device 1003 into the random access memory (RAM: Random Access Memory) 1004. In the RAM 1004, various programs and data required for the operation of the key person recognition device are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems may be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the key person recognition device to communicate with other devices wirelessly or wiredly to exchange data. Although the figure shows a key person recognition device with various systems, it should be understood that it is not required to implement or include all the systems shown. More or fewer systems may be alternatively implemented or included.

[0214] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program codes for executing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by a processing device 1001, the above-mentioned functions defined in the methods of the embodiments disclosed in the present application are executed.

[0215] The key person recognition device provided by the present application adopts the key person recognition method in the above-mentioned embodiment, and can solve the technical problem of low accuracy of the recognized key persons. Compared with the prior art, the beneficial effects of the key person recognition device provided by the present application are the same as those of the key person recognition method provided by the above-mentioned embodiment, and other technical features in the key person recognition device are the same as the features disclosed in the method of the previous embodiment, and will not be elaborated here.

[0216] It should be understood that each part disclosed in the present application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0217] As described above, only the specific embodiments of the present application are provided, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed in the present application can easily think of changes or substitutions, which should all be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0218] The present application provides a computer-readable storage medium with computer-readable program instructions (i.e., computer programs) stored thereon, and the computer-readable program instructions are used to execute the key person recognition method in the above-mentioned embodiment.

[0219] The computer-readable storage medium provided by this application can be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or components, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM) or flash memory, optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this embodiment, the computer-readable storage medium can be any tangible medium that contains or stores a program, which can be used by or in conjunction with an instruction execution system, device, or component. The program code contained on the computer-readable storage medium can be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0220] The above computer-readable storage medium can be included in the key person identification device; it can also exist separately without being assembled into the key person identification device.

[0221] The above computer-readable storage medium carries one or more programs. When the above one or more programs are executed by the key person identification device, the key person identification device is caused to: in response to a key person identification request, obtain user feature information;

[0222] store the user feature information into a preset social relationship graph through a hash function;

[0223] perform predictions on the social relationship graph through community network algorithms and machine learning algorithms to obtain a social relationship circle;

[0224] based on the social relationship circle, determine a target key person according to a preset feature discovery engine.

[0226] Computer program code for performing the operations of this application can be written in one or more programming languages or combinations thereof. The above-mentioned programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, by connecting through the Internet using an Internet service provider).

[0227] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of the code that contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0228] The modules involved in the embodiments described in this application can be implemented in software or in hardware. Among them, the name of the module does not constitute a limitation to the unit itself in some cases.

[0229] The readable storage medium provided in this application is a computer-readable storage medium. The computer-readable storage medium stores computer-readable program instructions (i.e., computer programs) for performing the above-mentioned key person recognition method, and can solve the technical problem of low accuracy in recognizing key persons. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the key person recognition method provided in the above embodiments, and will not be elaborated here.

[0230] The present application also provides a computer program product, including a computer program which, when executed by a processor, implements the steps of the key person recognition method as described above.

[0231] The computer program product provided by the present application can solve the technical problem of low accuracy in recognizing key persons. Compared with the prior art, the beneficial effects of the computer program product provided by the present application are the same as those of the key person recognition method provided in the above embodiments, and will not be elaborated here.

[0232] The above are only partial embodiments of the present application, and thus do not limit the patent scope of the present application. Any equivalent structural transformation made by using the content of the specification and drawings of the present application under the technical concept of the present application, or any direct / indirect application in other related technical fields, is included in the patent protection scope of the present application.

Claims

1. A method for identifying key figures, characterized in that, The described method includes: In response to a key person recognition request, obtaining user feature information; Storing the user feature information into a preset social relationship graph through a hash function; Predicting the social relationship graph through a community network algorithm and a machine learning algorithm to obtain a social relationship circle; Based on the social relationship circle, determining target key persons according to a preset feature discovery engine.

2. The method according to claim 1, wherein The step of storing the user feature information into a preset social relationship graph through a hash function includes: Obtaining user account information; Establishing a mapping relationship set between the user account information and the user feature information; Storing each element of the mapping relationship set into a preset social relationship graph through a hash function.

3. The method according to claim 1, characterized in that The step of predicting the social relationship graph through a community network algorithm and a machine learning algorithm to obtain a social relationship circle includes: Preliminarily dividing the social relationship graph through a community network algorithm, dividing user nodes into different communities according to the closeness of their social connections to obtain a preliminary social relationship circle; Based on the preliminary social relationship circle, extracting community feature information of each community; Inputting the community feature information into a preset relationship circle prediction model to predict and optimize the social relationship circle to obtain a social relationship circle.

4. The method according to claim 1, characterized in that, The feature discovery engine is obtained by fusing multiple models. The step of determining target key persons based on the social relationship circle and according to a preset feature discovery engine includes: Extracting features of user nodes in the social relationship circle to obtain user behavior features; Inputting the user behavior features into the feature discovery engine to evaluate and rank the user nodes to obtain a user node ranking result; Selecting target key persons from the user node ranking result according to a preset screening criterion.

5. The method according to claim 4, characterized in that, After the step of selecting target key persons from the user node ranking result according to a preset screening criterion includes: Obtaining social interaction data and business feedback data of the key persons; Analyzing and evaluating the social interaction data and the business feedback data to obtain an accuracy evaluation result; Adjusting the feature discovery engine based on the accuracy evaluation result.

6. The method according to claim 1, wherein After the step of obtaining user feature information in response to a key person recognition request includes: Judging the sensitive data type of the user feature information; Selecting a corresponding desensitization method to desensitize the user feature information according to the sensitive data type.

7. A key person recognition device, characterized in that, The device includes: An obtaining module, configured to obtain user feature information in response to a key person recognition request; A storage module, configured to store the user feature information into a preset social relationship graph through a hash function; A prediction module, configured to predict the social relationship graph through a community network algorithm and a machine learning algorithm to obtain a social relationship circle; A determination module, configured to determine target key persons based on the social relationship circle and according to a preset feature discovery engine.

8. A key person recognition device, characterized in that, The device includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, the computer program being configured to implement the steps of the key person recognition method according to any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the key person recognition method according to any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program, and when the computer program is executed by a processor, it implements the steps of the key person recognition method according to any one of claims 1 to 6.

Images