Information potential relation mining analysis method, device, equipment, medium and product

Abstract

The invention discloses an information potential relation mining analysis method, device and equipment, a medium and a product. The method comprises the following steps: screening out an information data set from a database, and constructing a user knowledge graph according to the information data set; determining a user group set according to the user knowledge graph, and obtaining target information corresponding to each user group; and determining a target site corresponding to each user group based on the target information of each user group, and determining a potential relationship corresponding to every two user groups according to the target information of each user group and the target site. According to the technical scheme, the information groups are automatically mined through the graph mining technology, the main information of the groups is obtained, then the potential relation between the groups is analyzed, the analysis efficiency is improved, the information processing efficiency and precision can be improved, and the requirements for real-time performance and automation are met.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of graph data and multi-model technology in artificial intelligence, and in particular, to a method, device, equipment, medium and product for mining and analyzing potential relationships of information. Background Art

[0002] In the context of globalization, hot events are often accompanied by a large number of public reports and comments. This information covers the information and positions of different users, forming a complex and diverse public opinion environment. The usual method of mining the analysis of groups related to hot events is to manually compile them into reports. Although this method can provide a certain depth and insights, it is limited by manpower and time costs, and it is difficult to achieve efficient processing of massive information. In addition, with the popularization of social media and digital communications, the amount of information has increased dramatically, making manual processing increasingly stretched. Therefore, there is an urgent need to develop a method that uses technical means to automatically mine and analyze the potential relationships between groups in information. Summary of the invention

[0003] The embodiments of the present invention provide a method, device, equipment, medium and product for mining and analyzing potential relationships of information, so as to improve the efficiency and accuracy of information processing and meet the requirements of real-time and automation.

[0004] According to one aspect of the present invention, a method for mining and analyzing potential information relationships is provided, comprising:

[0005] Filtering an information data set from a database, and constructing a user knowledge graph based on the information data set;

[0006] Determine a user group set according to the user knowledge graph, and obtain target information corresponding to each user group;

[0007] The target stance corresponding to each user group is determined based on the target information of each user group, and the potential relationship corresponding to every two user groups is determined based on the target information and the target stance of each user group.

[0008] According to another aspect of the present invention, there is provided a device for mining and analyzing potential information relationships, the device comprising:

[0009] A screening and construction module, used to screen out information data sets from the database and construct a user knowledge graph based on the information data sets;

[0010] A determination and acquisition module, used to determine a set of user groups according to the user knowledge graph, and to acquire target information corresponding to each user group;

[0011] The determination module is used to determine the target stance corresponding to each user group based on the target information of each user group, and to determine the potential relationship corresponding to every two user groups according to the target information and the target stance of each user group.

[0012] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0013] at least one processor; and

[0014] a memory communicatively connected to the at least one processor; wherein,

[0015] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the information potential relationship mining and analysis method described in any embodiment of the present invention.

[0016] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the information potential relationship mining and analysis method described in any embodiment of the present invention when executed.

[0017] According to another aspect of the present invention, an embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the information potential relationship mining and analysis method described in any embodiment of the present invention.

[0018] The embodiment of the present invention screens out relevant information data, calculates the relationship between users, constructs a user knowledge graph, uses the user knowledge graph to mine the group relationship between users, and then uses the information of the same group to obtain the main information of the current group. Finally, the position of the group and the potential relationship between groups are analyzed in combination with the main information of the group. Through the technical solution of the present invention, the manual compilation method of mining information groups is abandoned, and the information group is automatically mined through graph mining technology to obtain the main information of the group, and then the potential relationship between the groups is analyzed, which improves the analysis efficiency, can improve the efficiency and accuracy of information processing, and meet the needs of real-time and automation.

[0019] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 is a flow chart of a method for mining and analyzing potential information relationships in an embodiment of the present invention;

[0022] Figure 2 It is a structural schematic diagram of an information potential relationship mining and analysis device in an embodiment of the present invention;

[0023] Figure 3 It is a schematic diagram of the structure of an electronic device for implementing the information potential relationship mining and analysis method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and their meanings are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] It is understandable that before using the technical solutions disclosed in the embodiments of the present disclosure, the types, scope of use, usage scenarios, etc. of the personal information involved in the present disclosure should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations.

[0027] Embodiment 1

[0028] The analysis of potential relationships between groups is an issue that urgently needs to be understood. However, due to the large number of analytical dimensions involved, the challenges are also great. There are currently few analytical methods related to potential relationships between groups.

[0029] In the task of mining and analyzing potential relationships among groups, the traditional method is to rely on staff compilation. However, in the current era of massive data explosion, it is not enough to rely solely on staff compilation. When analyzing the position of certain information and whether there is a potential relationship between them, the staff collects and reviews them manually and summarizes them. In addition, it is difficult for inexperienced staff to make objective evaluations when refining and summarizing. This method can no longer fully meet the current needs of speed and accuracy in terms of efficiency. Moreover, manual compilation greatly increases the probability of missing important information when processing massive data. In addition, manual compilation often increases the possibility of subjective assumptions, and this method is also difficult to meet the needs of rapid changes in terms of efficiency.

[0030] If the analysis is done by the traditional manual compilation method, it is not only inefficient but also cannot meet the requirements of real-time and automation. Therefore, it is particularly important to develop a method that can automatically mine the potential relationship analysis of a group based on certain information.

[0031] Figure 1 is a flow chart of a method for mining and analyzing potential information relationships in an embodiment of the present invention. This embodiment is applicable to the case of mining and analyzing potential information relationships. The method can be executed by the device for mining and analyzing potential information relationships in an embodiment of the present invention. The device can be implemented in software and / or hardware, such as Figure 1 As shown, the method specifically comprises the following steps:

[0032] S101. Filter out an information data set from a database, and construct a user knowledge graph based on the information data set.

[0033] The method of the embodiment of the present invention can be applied to the information potential relationship mining analysis of hot events in the news. In this embodiment, the database can be all news events, and the information data set is screened out from the database, which can be understood as screening out specific hot events from the news events, that is, it can be understood that the information data set can be a set of news data screened from the news, in which different users publish information and positions on specific hot events.

[0034] It should be noted that the user knowledge graph can be a knowledge graph that describes the relationship between users by taking each information data as a node, calculating the relationship weight between two information nodes, and using the edge between the two information nodes as the edge.

[0035] In the specific implementation process, relevant information data of specific hot events are obtained by setting conditions for screening, and then the weight between each two information data is calculated to build the association relationship between users, thereby forming a detailed user knowledge graph.

[0036] S102: Determine a user group set based on the user knowledge graph, and obtain target information corresponding to each user group.

[0037] It should be noted that the user group set can be a set consisting of all groups in the entire user knowledge graph, which is obtained by dividing each information node in the user knowledge graph according to the group relationship between different users.

[0038] Among them, the target information can be the main information of each user group.

[0039] In this embodiment, the group relationship between users can be mined using the user knowledge graph based on the Leiden community discovery algorithm, and then the main information of the current group can be obtained based on the unsupervised clustering algorithm using the information of the same group.

[0040] S103: determining the target stance corresponding to each user group based on the target information of each user group, and determining the potential relationship corresponding to every two user groups according to the target information and the target stance of each user group.

[0041] The target position corresponding to the user group may be the position of the group on a hot event. For example, the target position corresponding to user group A may be, for example, a support position on a hot event, the target position corresponding to user group B may be, for example, an opposition position on a hot event, and the target position corresponding to user group C may be, for example, a neutral mediation position on a hot event.

[0042] It should be noted that the potential relationship may be the potential relationship between every two user groups with respect to a certain hot event.

[0043] Specifically, we can use the small sample learning ability of the large model and combine the main information of each user group to analyze the position of each user group, and then use the large model to analyze the potential relationship between each two user groups.

[0044] The embodiment of the present invention screens out relevant information data, calculates the relationship between users, constructs a user knowledge graph, uses the user knowledge graph to mine the group relationship between users, and then uses the information of the same group to obtain the main information of the current group. Finally, the position of the group and the potential relationship between groups are analyzed in combination with the main information of the group. Through the technical solution of the present invention, the manual compilation method of mining information groups is abandoned, and the information group is automatically mined through graph mining technology to obtain the main information of the group, and then the potential relationship between the groups is analyzed, which improves the analysis efficiency, can improve the efficiency and accuracy of information processing, and meet the needs of real-time and automation.

[0045] Optionally, an information data set is screened out from the database, and a user knowledge graph is constructed based on the information data set, including:

[0046] According to preset keywords, preset time range and preset information type, the information data set is filtered out from the database.

[0047] In this embodiment, the preset keywords may be, for example, filtering words pre-set by the user according to actual needs, the preset time range may be a filtering time range set by the user, for example, filtering according to the time when an event occurs or the time when information is released, and the preset information type may be the type information of events filtered from the news set by the user, for example, sports, entertainment, life, technology, etc.

[0048] Specifically, multiple screening conditions such as specific keywords, time ranges, and event types can be pre-set to screen out information data sets from the database. In actual operation, the information data in the screened information data set can be repeatedly filtered.

[0049] A keyword set is extracted from each information data in the information data set.

[0050] In this embodiment, the keyword set may be a set of keywords extracted from information data and preset by a user. The keyword set may include, for example, a country name, an organization name, an institution name, and a user name.

[0051] Specifically, for each information data in the information data set, each keyword in the keyword set is extracted in turn, for example, information such as the country name, organization name, institution name, and user name contained in each information data is extracted.

[0052] If the first keyword and the second keyword corresponding to two information data are the same, the relationship weight between the two information data is set to the first weight.

[0053] In this embodiment, the first keyword may be, for example, a country name, and the second keyword may be, for example, an organization name. The first keyword and the second keyword corresponding to the two information data are the same, that is, it can be understood that the contents of the two information data are relevant information about the same country and the same organization.

[0054] The first weight may be preset by the user according to actual needs or experience, and is used to represent the relationship weight between two information data having the same first keyword and second keyword.

[0055] Specifically, a judgment is made for the keywords extracted from every two pieces of information data in the information data set. If the country names and organization names corresponding to the two pieces of information data are the same, the weight of the relationship between the two pieces of information data is set to 1.

[0056] If there are two information data whose corresponding first keywords are the same and whose second keywords are different, the relationship weight between the two information data is set to the second weight.

[0057] The second weight may be preset by the user according to actual needs or experience, and is used to represent the relationship weight between two pieces of information data with the same first keyword but different second keywords. Preferably, the second weight may be set to 0.7.

[0058] Specifically, if the country names corresponding to two information data are the same but the organization names are different, that is, the contents of the two information data are related to the same country but not the same organization, the relationship weight between the two information data is set to 0.7.

[0059] If there are two information data whose corresponding first keywords and second keywords are different, the two information data are vectorized respectively, and the relationship weight between the two information data is calculated based on the two vectorized information data. If the calculated relationship weight is greater than the weight threshold, the relationship weight between the two information data is set to the calculated relationship weight.

[0060] The weight threshold may be a threshold preset by the user according to actual needs or experience. Preferably, the weight threshold may be set to 0.68, for example.

[0061] In this embodiment, if the country names and organization names corresponding to the two information data are different, the two information data are vectorized into V 信息列表 , and then respectively for the two vectorized information data V 信息列表 Perform mean pooling: Then the similarity weight value between the two information data is calculated by cosine similarity. The calculation method of cosine similarity can be expressed as:

[0062]

[0063] Among them, V X信息均值 and V Y信息均值 Respectively represent the one-dimensional vectors after the mean pooling of the current two information data. Finally, the similarity weight value between the two information data is obtained, and it is determined whether the weight value is greater than 0.68. If so, the relationship weight between the two is set to the weight value; otherwise, no relationship is set between the two.

[0064] Build a user knowledge graph based on the relationship weight between every two pieces of information data.

[0065] Specifically, after calculating the relationship weight between every two information data, each information data is used as an information node, and the relationship weight between every two information data is used as the edge between every two information nodes to construct a user knowledge graph.

[0066] Optionally, determine a user group set based on the user knowledge graph, and obtain target information corresponding to each user group, including:

[0067] Based on the community discovery algorithm, each information data is divided into each corresponding modularity gain in each to-be-determined group, and each information data is merged to obtain a user group set.

[0068] After constructing the user knowledge graph, it is necessary to mine the potential group relationships between users in the graph network under the hot event. In the past, when using community discovery algorithms to find community information, they are often sensitive to density differences in the input data, which may result in the discovery of any poorly connected communities. In some cases, there may be problems with over-segmentation, which will lead to large errors. The embodiment of the present invention introduces a method based on the Leiden community discovery algorithm combined with the user knowledge graph to mine potential information groups. The Leiden community discovery algorithm is an improved algorithm of Louvain. The Leiden community discovery algorithm introduces a more sophisticated termination condition, which can avoid the problem of over-segmentation in some cases, thereby generating a higher quality information group structure. By splitting the group community in a specific way, the Leiden community discovery algorithm ensures good connections between group communities and optimal characteristics within group communities, thereby generating higher quality group communities.

[0069] The Leiden community discovery algorithm is an unsupervised community discovery algorithm based on modularity optimization, which is developed and optimized based on the Louvain algorithm. Among them, modularity is an important indicator for testing the quality of group division. The modularity Q calculation method can flexibly adjust the number and size of groups. Specifically, the modularity calculation formula can be expressed as follows:

[0070]

[0071] Among them, m is the sum of the relationship weights of all edges in the network graph of the user knowledge graph, k i is the sum of the relationship weights of all edges pointing to information node i, A ij is the relationship weight between information nodes i and j, c i is the group to which information node i is assigned, and γ is the adjustment parameter, which is positively correlated with the number of groups.

[0072] In practice, γ is a resolution parameter that adjusts the density of connections within and between groups. γ is positively correlated with the number of reaction communities. The size and rationality of a group can be adjusted and analyzed with the help of γ. For a group network, if a group is too large, the group consistency of the users included in the group may not be pure, which is not conducive to the subsequent group position analysis; conversely, if a group is too small, users who should be in one group may be divided into multiple groups, which is not conducive to the analysis of potential relationships between groups. Therefore, the standard for the value of γ should be to make each group contain a complete, clear and uncoupled group mechanism. In particular, when c i =c j When δ(c i ,c j )=1, otherwise it is 0. The range of modularity is The higher the modularity, the more stable the population structure and the more obvious the modular characteristics.

[0073] In actual operation, each information node in the graph network of the user knowledge graph is assigned to a separate group, and then these groups are gradually merged by calculating the modularity gain ΔQ obtained by moving the information node i to the group c. The calculation method of the modularity gain ΔQ can be expressed as:

[0074]

[0075] Among them, k i,in is the sum of the weights of the information node edges in group c, k i is the weighted total weight of information node i, ∑ total is the sum of the relationship weights of all edges connected to information node i. The group is continuously adjusted and optimized until ΔQ no longer changes, which means that the modularity Q of the current group reaches the maximum value, and a roughly divided group is obtained.

[0076] After obtaining the roughly divided groups as mentioned above, it is necessary to perform internal subdivision of the groups. For each group, the subdivision operation can only be implemented internally, and the further subdivided groups are called subgroups. Each information node in each group is initially assigned to a partition, and then the random neighborhood Louvain algorithm is used to detect the subgroups of these partitions. It is particularly important to note that the random neighborhood Louvain algorithm allows the addition of information nodes in the subgroup to only be within the upper-level group. After implementing the random neighborhood Louvain algorithm, the roughly divided groups are converted into more smaller subgroups.

[0077] The obtained subgroups are then regarded as new "nodes" and the above steps are iterated until there is no further improvement and the entire network containing the group is obtained. Specifically, the user group set can be expressed as:

[0078] G={G1,G2,…,G n}={(V1,E1),(V2,E2),…,(V n ,E n )};

[0079] Among them, G is the user group set, n is the total number of groups included in the user group set, G n For each user group, V n represents the number of information nodes contained in the nth user group, E n Represents the number of edges contained in the nth user group.

[0080] For each user group, the central probability value corresponding to each information data is obtained based on the unsupervised clustering algorithm.

[0081] Each user may have a lot of information, and the meaning of the expression may also involve many directions, which requires extracting the main core information involved in each user. Therefore, the embodiment of the present invention introduces an unsupervised clustering algorithm to automatically classify information with similar semantic features into the same group by combining the semantic vector of the information, and obtains the most important information of various groups to form the main information. This embodiment does not limit the specific method and process of the unsupervised clustering algorithm.

[0082] In this embodiment, a plurality of information clusters are obtained from a given list of information data in a certain user group through an unsupervised clustering algorithm, and each information data can obtain the cluster label to which it belongs and the center probability value of the cluster to which it belongs.

[0083] The information data with the largest central probability value in each user group is used as the target information corresponding to the user group.

[0084] Specifically, all information data under the current information cluster are sorted according to the central probability value of the cluster to which they belong, and the information data with the largest central probability value is used as the main information corresponding to the current user group.

[0085] Optionally, determining a target position corresponding to each user group based on the target information of each user group includes:

[0086] The target information of each user group is input into the group stance analysis network model to obtain the target stance corresponding to each user group.

[0087] In this embodiment, the group stance analysis network model may be a model for mining the potential stance of a group on a hot event and its stance information. Preferably, the group stance analysis network model may be a large model, which is not specifically limited in this embodiment.

[0088] The group stance analysis network model is obtained by iteratively training a neural network model using a group stance analysis data set. The group stance analysis data set includes: information input samples and stance output samples.

[0089] It should be noted that the group stance analysis dataset may be information from several groups and their corresponding stances, annotated by the user and used to train the neural network model. The information input samples may be information selected from several groups, and the stance output samples may be the stances corresponding to each information input sample. Preferably, both the information input samples and the stance output samples may be annotated in advance by the user.

[0090] Specifically, the main information of each user group is obtained, and then a small amount of group stance analysis data sets are labeled. The large model is trained to obtain a group stance analysis network model. Then, the target information of each user group is input into the trained group stance analysis network model. The large model's few-sample learning ability and contextual learning ability are used to mine the potential stance and stance information of each user group on current hot events.

[0091] Optionally, the potential relationship between every two user groups is determined based on the target information and target stance of each user group, including:

[0092] The target positions of every two user groups are obtained, and the position consistency results corresponding to every two user groups are determined.

[0093] It should be noted that the position consistency result can be whether the target positions of the two user groups are consistent, that is, it can be understood as whether there is an intersection or consistency in the words corresponding to the target positions of the two user groups. For example, if the target position of user group A is support and the target position of user group B is neutral, then the position consistency result corresponding to user group A and user group B is inconsistent; if the target position of user group C is opposition and the target position of user group D is strong opposition, then the position consistency result corresponding to user group C and user group D is consistent.

[0094] Specifically, the target positions of each two user groups are similarly judged to determine whether the target positions of each two user groups are the same or have an intersection. For example, if the target positions of the two user groups are the same or have an intersection, it can be determined that the positions of the two user groups are consistent; if the target positions of the two user groups are different or have no intersection, it can be determined that the positions of the two user groups are inconsistent.

[0095] The target information of every two user groups is input into the network model for judging the information relevance between groups, and the corresponding relevance judgment results of every two user groups are obtained.

[0096] In this embodiment, the network model for judging the information relevance between groups may be a model for judging whether two user groups are related. Preferably, the network model for judging the information relevance between groups may be a large model, and this embodiment does not specifically limit the large model.

[0097] The network model for judging the information relevance between groups is obtained by iteratively training a neural network model using a group information data set, and the group information data set includes: information input samples corresponding to at least two user groups and relevance judgment result output samples.

[0098] It should be noted that the group information data set can be user-annotated, and used to train the neural network model, and includes information from several different user groups and their corresponding relevance judgment results. Among them, the information input samples corresponding to at least two user groups can be information selected from several different user groups respectively, and the relevance judgment result output samples can be the relevance judgment results between the information input samples corresponding to every two user groups. Preferably, the information input samples corresponding to at least two user groups and the relevance judgment result output samples can both be annotated in advance by the user.

[0099] It should be noted that the result of the correlation determination may be that the information of the two user groups is correlated, or the result of the correlation determination may be that the information of the two user groups is not correlated.

[0100] Specifically, the target information of every two user groups is input into the network model for judging the information relevance between groups, and the relevance judgment results corresponding to every two user groups are obtained by utilizing the small sample learning ability and context learning ability of the large model.

[0101] If the position consistency results corresponding to two user groups are consistent and the correlation judgment result is that there is a correlation, then it is determined that the potential relationship corresponding to the two user groups is consistent.

[0102] Specifically, if the target positions of two user groups are the same or overlap, and the information of the two user groups is related, then the potential relationship corresponding to the two user groups is determined to be consistent; otherwise, the potential relationship corresponding to the two user groups is determined to be inconsistent.

[0103] Optionally, after determining that the potential relationships between the two user groups are consistent, the method further includes:

[0104] The target information corresponding to two user groups with the same potential relationship is input into the group potential relationship analysis network model to obtain the group potential relationship probability.

[0105] In this embodiment, the group potential relationship analysis network model may be a model for determining whether the target stance identification analysis results corresponding to two user groups with consistent potential relationships are valid.

[0106] The group potential relationship analysis network model is obtained by iteratively training a neural network model using a group potential relationship analysis data set. The group potential relationship analysis data set includes: target information input samples corresponding to at least two user groups and group potential relationship probability output samples.

[0107] It should be noted that the group potential relationship analysis data set can be user-annotated, and used to train the neural network model, and includes information from several different user groups and their corresponding group potential relationship probabilities. Among them, the target information input samples corresponding to at least two user groups can be information selected from several different user groups respectively, and the group potential relationship probability output samples can be the group potential relationship probability between the target information input samples corresponding to every two user groups. Preferably, the target information input samples corresponding to at least two user groups and the group potential relationship probability output samples can be annotated in advance by the user.

[0108] It should be noted that the group potential relationship probability may be a probability value of whether a potential relationship analysis result between target information corresponding to two user groups with consistent potential relationships is valid.

[0109] Specifically, after the two user groups determine that the potential relationship is consistent, the main information of the two user groups is labeled, a small number of group potential relationship analysis data sets are constructed, the neural network model is trained to obtain the group potential relationship analysis network model, and then the target information corresponding to the two user groups with consistent potential relationships is input into the group potential relationship analysis network model to obtain the group potential relationship probability.

[0110] If the probability of a potential relationship between groups is greater than a preset threshold, the potential relationship is deemed valid.

[0111] The preset threshold may be a threshold preset by the user according to actual needs or experience. Preferably, the preset threshold may be set to 85%, for example.

[0112] Specifically, it is determined whether the group potential relationship probability output by the group potential relationship analysis network model is greater than a set threshold of 85%. If it is greater than the threshold, the analysis result of the potential relationship is deemed valid; otherwise, the analysis result is deemed invalid.

[0113] The technical solution of the embodiment of the present invention filters out relevant information data of a hot event according to multiple conditions, calculates the relationship between users through relevant semantics and rules, and constructs a user knowledge graph; then introduces a Leiden community discovery algorithm combined with a user knowledge graph to mine potential information groups. First, the manual compilation method of mining information groups is abandoned, and information groups are automatically mined through graph mining technology, which improves the analysis efficiency. Secondly, the Leiden community discovery algorithm introduces a more sophisticated termination condition when continuously optimizing the modularity to find the best division of the network, which can avoid the problem of over-segmentation in some cases, thereby generating a higher quality information group structure; then, an unsupervised clustering algorithm is introduced in combination with the semantic vector of information to better and faster obtain the main information of a group; and the small sample learning ability of the large model is introduced to analyze the position of the group and the potential relationship between groups, and the context understanding ability of the large model can be used to more quickly and objectively extract more reasonable analysis results. Through the technical solution of the present invention, the manual compilation method of mining information groups is abandoned, and the information groups are automatically mined through graph mining technology to obtain the main information of the groups, and then the potential relationships between the groups are analyzed, which improves the analysis efficiency and can improve the efficiency and accuracy of information processing to meet the needs of real-time and automation.

[0114] Embodiment 2

[0115] Figure 2Schematic diagram of the structure of an information potential relationship mining and analysis device in an embodiment of the present invention. This embodiment can be applied to the situation of information potential relationship mining and analysis based on news hot events. The device can be implemented in software and / or hardware. The device can be integrated in any device that provides the function of information potential relationship mining and analysis, such as Figure 2 As shown, the information potential relationship mining and analysis device specifically includes: a screening and construction module 201, a determination and acquisition module 202 and a determination module 203.

[0116] The screening and construction module 201 is used to screen out information data sets from the database and construct a user knowledge graph based on the information data sets;

[0117] A determination and acquisition module 202 is used to determine a set of user groups according to the user knowledge graph and obtain target information corresponding to each user group;

[0118] The determination module 203 is used to determine the target stance corresponding to each user group based on the target information of each user group, and determine the potential relationship corresponding to every two user groups according to the target information and the target stance of each user group.

[0119] Optionally, the screening and construction module 201 is specifically used for:

[0120] Filtering information data sets from a database according to preset keywords, preset time ranges, and preset information types;

[0121] Extracting a keyword set from each information data in the information data set;

[0122] If the first keyword and the second keyword corresponding to two pieces of information data are the same, the relationship weight between the two pieces of information data is set to the first weight;

[0123] If there are two pieces of information data corresponding to the same first keyword and different second keywords, the relationship weight between the two pieces of information data is set to a second weight;

[0124] If there are two pieces of information data corresponding to the first keyword and the second keyword that are different, the two pieces of information data are respectively vectorized, and the relationship weight between the two pieces of information data is calculated according to the two pieces of information data after vectorization, and if the calculated relationship weight is greater than the weight threshold, the relationship weight between the two pieces of information data is set to the calculated relationship weight;

[0125] Build a user knowledge graph based on the relationship weight between every two pieces of information data.

[0126] Optionally, the determining and obtaining module 202 is specifically used for:

[0127] Based on the community discovery algorithm, each piece of information data is merged according to the corresponding modularity gain of each group to be determined, and a user group set is obtained;

[0128] For each user group, the central probability value corresponding to each information data is obtained based on the unsupervised clustering algorithm;

[0129] The information data with the largest central probability value in each user group is used as the target information corresponding to the user group.

[0130] Optionally, the determining module 203 is specifically configured to:

[0131] The target information of each user group is input into a group stance analysis network model to obtain the target stance corresponding to each user group. The group stance analysis network model is obtained by iteratively training a neural network model with a group stance analysis data set. The group stance analysis data set includes: information input samples and stance output samples.

[0132] Optionally, the determining module 203 is specifically configured to:

[0133] Obtain the target positions of every two user groups, and determine the position consistency results corresponding to every two user groups;

[0134] Inputting the target information of every two user groups into the inter-group information relevance judgment network model to obtain the relevance judgment results corresponding to every two user groups, wherein the inter-group information relevance judgment network model is obtained by iteratively training the neural network model with a group information data set, wherein the group information data set includes: information input samples corresponding to at least two user groups and relevance judgment result output samples;

[0135] If the position consistency results corresponding to two user groups are consistent and the correlation judgment result is that there is a correlation, then it is determined that the potential relationship corresponding to the two user groups is consistent.

[0136] Optionally, the device is also used for:

[0137] Inputting target information corresponding to two user groups with consistent potential relationships into a group potential relationship analysis network model to obtain group potential relationship probability, wherein the group potential relationship analysis network model is obtained by iteratively training a neural network model with a group potential relationship analysis data set, wherein the group potential relationship analysis data set includes: target information input samples corresponding to at least two user groups and group potential relationship probability output samples;

[0138] If the probability of the potential relationship between the groups is greater than a preset threshold, the potential relationship is deemed valid.

[0139] The above-mentioned product can execute the information potential relationship mining and analysis method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0140] Embodiment 3

[0141] Figure 3 A schematic diagram of the structure of an electronic device 30 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0142] like Figure 3 As shown, the electronic device 30 includes at least one processor 31, and a memory connected to the at least one processor 31, such as a read-only memory (ROM) 32, a random access memory (RAM) 33, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 31 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 32 or the computer program loaded from the storage unit 38 to the random access memory (RAM) 33. In the RAM 33, various programs and data required for the operation of the electronic device 30 can also be stored. The processor 31, the ROM 32, and the RAM 33 are connected to each other through a bus 34. An input / output (I / O) interface 35 is also connected to the bus 34.

[0143] A number of components in the electronic device 30 are connected to the I / O interface 35, including: an input unit 36, such as a keyboard, a mouse, etc.; an output unit 37, such as various types of displays, speakers, etc.; a storage unit 38, such as a disk, an optical disk, etc.; and a communication unit 39, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 39 allows the electronic device 30 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0144] The processor 31 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 31 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 31 performs the various methods and processes described above, such as the information potential relationship mining analysis method:

[0145] Filtering an information data set from a database, and constructing a user knowledge graph based on the information data set;

[0146] Determine a user group set according to the user knowledge graph, and obtain target information corresponding to each user group;

[0147] The target stance corresponding to each user group is determined based on the target information of each user group, and the potential relationship corresponding to every two user groups is determined based on the target information and the target stance of each user group.

[0148] In some embodiments, the information potential relationship mining analysis method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 38. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 30 via the ROM 32 and / or the communication unit 39. When the computer program is loaded into the RAM 33 and executed by the processor 31, one or more steps of the information potential relationship mining analysis method described above can be performed. Alternatively, in other embodiments, the processor 31 can be configured to execute the information potential relationship mining analysis method in any other appropriate manner (for example, by means of firmware).

[0149] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0150] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0151] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0152] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0153] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0154] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0155] In one embodiment, the embodiment of the present invention further includes a computer program product, the computer program product includes a computer program, and when the computer program is executed by a processor, it implements the information potential relationship mining and analysis method of any embodiment of the present invention.

[0156] In the process of implementation, the computer program product can be written in one or more programming languages ​​or a combination thereof to perform the computer program code of the present invention, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a separate 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 can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0157] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0158] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for mining and analyzing potential information relationships, characterized in that: include: Filtering an information data set from a database, and constructing a user knowledge graph based on the information data set; Determine a user group set according to the user knowledge graph, and obtain target information corresponding to each user group; The target stance corresponding to each user group is determined based on the target information of each user group, and the potential relationship corresponding to every two user groups is determined based on the target information and the target stance of each user group.

2. The method according to claim 1, characterized in that Filtering an information data set from a database and constructing a user knowledge graph based on the information data set includes: Filtering information data sets from a database according to preset keywords, preset time ranges, and preset information types; Extracting a keyword set from each information data in the information data set; If the first keyword and the second keyword corresponding to two pieces of information data are the same, the relationship weight between the two pieces of information data is set to the first weight; If there are two pieces of information data corresponding to the same first keyword and different second keywords, the relationship weight between the two pieces of information data is set to a second weight; If there are two pieces of information data corresponding to the first keyword and the second keyword that are different, the two pieces of information data are respectively vectorized, and the relationship weight between the two pieces of information data is calculated according to the two pieces of information data after vectorization, and if the calculated relationship weight is greater than the weight threshold, the relationship weight between the two pieces of information data is set to the calculated relationship weight; Build a user knowledge graph based on the relationship weight between every two pieces of information data.

3. The method according to claim 1, characterized in that Determine a user group set according to the user knowledge graph, and obtain target information corresponding to each user group, including: Based on the community discovery algorithm, each piece of information data is merged according to the corresponding modularity gain of each group to be determined, and a user group set is obtained; For each user group, the central probability value corresponding to each information data is obtained based on the unsupervised clustering algorithm; The information data with the largest central probability value in each user group is used as the target information corresponding to the user group.

4. The method according to claim 1, characterized in that: Determine the target position corresponding to each user group based on the target information of each user group, including: The target information of each user group is input into a group stance analysis network model to obtain the target stance corresponding to each user group. The group stance analysis network model is obtained by iteratively training a neural network model with a group stance analysis data set. The group stance analysis data set includes: information input samples and stance output samples.

5. The method according to claim 1, characterized in that Determine the potential relationship between every two user groups based on the target information and target stance of each user group, including: Obtain the target positions of every two user groups, and determine the position consistency results corresponding to every two user groups; Inputting the target information of every two user groups into the inter-group information relevance judgment network model to obtain the relevance judgment results corresponding to every two user groups, wherein the inter-group information relevance judgment network model is obtained by iteratively training the neural network model with a group information data set, wherein the group information data set includes: information input samples corresponding to at least two user groups and relevance judgment result output samples; If the position consistency results corresponding to two user groups are consistent and the correlation judgment result is that there is a correlation, then it is determined that the potential relationship corresponding to the two user groups is consistent.

6. The method according to claim 5, characterized in that After determining that the potential relationship between the two user groups is consistent, it also includes: Inputting target information corresponding to two user groups with consistent potential relationships into a group potential relationship analysis network model to obtain group potential relationship probability, wherein the group potential relationship analysis network model is obtained by iteratively training a neural network model with a group potential relationship analysis data set, wherein the group potential relationship analysis data set includes: target information input samples corresponding to at least two user groups and group potential relationship probability output samples; If the probability of the potential relationship between the groups is greater than a preset threshold, the potential relationship is deemed valid.

7. An information potential relationship mining and analysis device, characterized in that: include: A screening and construction module, used to screen out information data sets from the database and construct a user knowledge graph based on the information data sets; A determination and acquisition module, used to determine a set of user groups according to the user knowledge graph, and to acquire target information corresponding to each user group; The determination module is used to determine the target stance corresponding to each user group based on the target information of each user group, and to determine the potential relationship corresponding to every two user groups according to the target information and the target stance of each user group.

8. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the information potential relationship mining and analysis method described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the information potential relationship mining and analysis method according to any one of claims 1 to 6 when executed.

10. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the information potential relationship mining and analysis method according to any one of claims 1 to 6.

Images