A social user topic analysis method and system based on semantics and word expansion

Abstract

The present invention discloses a social user topic analysis method and system based on semantics and word expansion, which relates to the technical field of network user information assessment, and solves the problem that there is likely to be no contextual association between user posting information in the existing topic analysis method. The technical solution is: a short text topic model based on semantics and word expansion is proposed, which not only integrates the semantic information of the short text, but also expands word pairs through external corpus knowledge, while solving the short text sparsity problem, enhancing the topic tendency of the text; on top of the short text topic model based on semantics and word expansion, a topic weighting method based on the amount of interaction between users is adopted to distinguish the importance of posts, thereby calculating the topic distribution of users. Compared with the traditional topic feature analysis method, the social user topic analysis method of the present invention is more reasonable.

Description

Technical Field

[0001] The present invention relates to the technical field of network user information evaluation, and more specifically, to a social user topic analysis method and system based on semantics and word expansion. Background Art

[0002] With the continuous development of the Internet, short text data on the network has exploded, and people can publish various information on various social platforms. Therefore, how to analyze user characteristics from social platforms is very valuable and practical. Among them, the theme characteristics of social users are one of the research focuses. At present, most of the information dissemination on major social platforms such as Weibo, Twitter, WeChat, etc. is in the form of short text. Short text data is different from long text data at the chapter level. Short text data has different language rules. Users are the main body of social platforms, and each user can publish thousands of short text messages. For enterprises, analyzing the theme information of users can allow enterprises to propose targeted solutions to improve user experience, which has certain commercial value. For scholars, analyzing the theme distribution of users can be used as a social science research result.

[0003] At present, the common method for extracting topic features from users on social platforms is to integrate all posts of the same user into a whole, making it a long document, and then input the long document into the topic model to obtain the topic distribution of the document, and regard this distribution as the topic distribution of the user. This can solve the problem of sparse short texts and also obtain "user-level" topic features. However, there may be no contextual association between the posts of the user, and they may not even be about the same topic. Therefore, it is unreasonable to simply merge all posts of the same user into a pseudo-document.

[0004] Therefore, how to make the analyzed topic distribution characteristics of users more reasonable is a problem that needs to be solved urgently. Summary of the invention

[0005] The purpose of the present invention is to provide a social user topic analysis method and system based on semantics and word expansion; it solves the problem that there is no contextual association between user postings with topic distribution characteristics obtained by the topic analysis method of the prior art, and even the problems of not describing things of the same topic; the present invention improves the traditional short text topic model and proposes a short text topic model based on semantics and word expansion, which performs topic analysis on each social posting information to obtain a topic distribution matrix, and then calculates the weight matrix of the posting information, that is, the importance of the posting information, according to the interactive information of each posting information, and finally calculates the user's topic distribution using the topic distribution matrix and the weight matrix. Compared with the user topic analysis method of the prior art that directly constructs pseudo-documents, the social user topic analysis method based on semantics and word expansion of the present invention is more advantageous.

[0006] The above technical objectives of the present invention are achieved through the following technical solutions:

[0007] In a first aspect, a social user topic analysis method based on semantics and word expansion is provided, comprising the following steps:

[0008] Obtaining the user's posting information and the interactive information generated by the posting information in the social networking platform;

[0009] Performing a preprocessing operation on the posting information to obtain text data of the posting information;

[0010] Constructing a short text topic model based on semantics and word expansion according to the text data;

[0011] Inputting the posting information into the short text topic model for calculation to obtain a plurality of first topic distribution matrices of the posting information;

[0012] Calculate a weight matrix for each posting information of the user according to the interactive information;

[0013] Performing weighted processing on the plurality of first topic distribution matrices according to the weight matrix to obtain a plurality of second topic distribution matrices;

[0014] Vector merging and normalization processing are performed on the multiple second topic distribution matrices to obtain the topic distribution of the posting information of the user.

[0015] Compared with the prior art, the present invention improves the traditional short text topic model and proposes a short text topic model based on semantics and word expansion. The model performs topic analysis on each social post information to obtain a topic distribution matrix, and then calculates the weight matrix of the post information according to the interactive information of each post information, that is, the importance of the post information. Finally, the topic distribution matrix and the weight matrix are used to calculate the user's topic distribution. Compared with the user topic analysis method of directly constructing pseudo-documents in the prior art, the social user topic analysis method based on semantics and word expansion of the present invention is more advantageous.

[0016] Furthermore, the preprocessing operation includes performing a word segmentation operation on the post information, removing stop words from the post information, and removing interference symbols from the post information.

[0017] Furthermore, the steps of constructing a short text topic model based on semantics and word expansion according to the text data are as follows:

[0018] Extracting a first word pair with semantic association in the context of the text data by using a semantic dependency analysis method;

[0019] The keywords to be expanded in the text data are obtained, the keywords to be expanded are input into an external corpus, and the correlation between the keywords to be expanded and words in the external corpus is calculated using point mutual information.

[0020] Furthermore, a judgment threshold of the correlation is set, and if the obtained correlation is greater than the judgment threshold, the keyword to be expanded and the word in the external corpus are combined into a second word pair.

[0021] Furthermore, all vectors in the multiple second topic distribution matrices are merged to obtain the user's topic distribution feature vector, and the topic distribution feature vector is normalized using a softmax function to obtain the topic distribution of the user's posting information.

[0022] In a second aspect, a social user topic analysis system based on semantics and word expansion is provided, including:

[0023] An information acquisition unit, used to acquire the user's posting information and the interactive information generated by the posting information in the social network platform;

[0024] A preprocessing unit, used to perform a preprocessing operation on the posting information to obtain text data of the posting information;

[0025] A model building unit, used for building a short text topic model based on semantics and word expansion according to the text data;

[0026] A first calculation unit, configured to input the post information into the short text topic model for calculation to obtain a plurality of first topic distribution matrices of the post information;

[0027] A second calculation unit, used for calculating a weight matrix of each posting information of the user according to the interaction information;

[0028] A weighted processing unit, used for performing weighted processing on the plurality of first topic distribution matrices according to the weight matrix to obtain a plurality of second topic distribution matrices;

[0029] The merging processing unit is used to perform vector merging and normalization processing on the multiple second topic distribution matrices to obtain the topic distribution of the user's posting information.

[0030] Further, the preprocessing unit includes a word segmentation unit and a removal unit;

[0031] The word segmentation unit is used to perform word segmentation operations on the post information, and the removal unit is used to remove stop words from the post information and remove interference symbols in the post information.

[0032] Further, the model building unit includes a semantic extraction unit and a word expansion unit;

[0033] The semantic extraction unit is used to extract the first word pair with semantic association in the context of the text data by using a semantic dependency analysis method;

[0034] The word expansion unit is used to obtain the keywords to be expanded from the text data, input the keywords to be expanded into an external corpus, and calculate the correlation between the keywords to be expanded and the words in the external corpus using point mutual information.

[0035] Furthermore, the word expansion unit also includes a judgment unit;

[0036] The judgment unit is used to set a judgment threshold of the correlation. If the obtained correlation is greater than the judgment threshold, the keyword to be expanded and the word in the external corpus are combined into a second word pair.

[0037] Furthermore, the merging processing unit also includes a vector merging unit and a processing unit;

[0038] The vector merging unit is used to merge all the vectors in the plurality of second topic distribution matrices to obtain a topic distribution feature vector of the user;

[0039] The processing unit is used to normalize the topic distribution feature vector using a softmax function to obtain the topic distribution of the user's posting information.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] 1. The analysis method of the present invention proposes a short text topic model based on semantics and word expansion. Based on the traditional short text topic model, this model not only integrates the semantic information of the short text, but also expands word pairs through an external corpus knowledge base, thereby solving the sparsity problem of short texts and enhancing the topic tendency of the text.

[0042] 2. Based on the short text topic model based on semantics and word expansion, the present invention adopts a topic weighting method based on the interactive information between user posts to distinguish the importance of posts, thereby calculating the user's topic distribution. Compared with the traditional topic feature calculation method, it is more reasonable. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0044] Figure 1 A schematic diagram of a process flow of a social user topic analysis method provided by an embodiment of the present invention;

[0045] Figure 2 A schematic diagram for obtaining word pairs of short text topic models in the prior art;

[0046] Figure 3 A schematic diagram of word pair acquisition of a short text topic model based on semantics and word expansion provided by an embodiment of the present invention;

[0047] Figure 4 A schematic diagram of a process for obtaining user topic distribution according to an embodiment of the present invention;

[0048] Figure 5 A schematic flow chart of a method for constructing a short text topic model based on semantics and word expansion provided by an embodiment of the present invention;

[0049] Figure 6 A schematic diagram of the structure of a social user topic analysis system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0051] It should be noted that the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0052] Embodiment 1:

[0053] like Figure 1 As shown, the first embodiment of the present application provides a social user topic analysis method based on semantics and word expansion, comprising the following steps:

[0054] S1, obtaining the user's posting information and the interactive information generated by the posting information in the social network platform;

[0055] The user's posting information includes the user's fixed personal information, such as user ID, and the information of his original works, such as article ID, posting content, etc. The posting information mainly refers to short text information. The interactive behavior generated with each posting information is interactive information, including the number of reposts, the number of likes, the number of favorites, the number of comments, the number of follows, and the number of shares. To analyze the topic distribution of posts by social network users, this application is based on all posting information of social network users, as well as the interactive information that generates interactive behavior with each posting information.

[0056] S2, performing a preprocessing operation on the posting information to obtain text data of the posting information;

[0057] Performing preprocessing operations on posted information to obtain text data of the posted information is a conventional technical means in the art, and therefore no redundant description will be given.

[0058] S3, constructing a short text topic model based on semantics and word expansion according to the text data;

[0059] Specifically, Figure 2 As shown, Figure 2That is, the short text topic model (BTM model for short) in the prior art is used to obtain word pair information of short texts, but the semantic association of the word pairs in the context of the short text is ignored when the sliding window method is used to obtain the word pair information. Therefore, the short text topic model based on semantics and word expansion of the present application adopts a semantic dependency method to extract word pairs in the text, so that the BTM model can integrate the semantic information of the context. In the prior art, the user's topic features are mainly extracted by constructing pseudo-documents, which can solve the problem of short text sparsity. Therefore, in order to enable the short text topic model proposed in the present application to solve the problem of short text sparsity, a word pair expansion method based on word association (i.e., a short text topic model with word expansion) is proposed, so that the BTM model can solve the problem of short text sparsity.

[0060] S4, inputting the posting information into the short text topic model for calculation to obtain a plurality of first topic distribution matrices of the posting information;

[0061] By inputting all the user's posting information into the short text topic model based on semantics and word expansion, the first topic distribution matrix of each posting information can be obtained. The topic distribution of all the posting information of each user is actually a D*K-dimensional first topic distribution matrix, where D represents all the user's posting information and K represents the set number of topics. The D*K-dimensional first topic distribution matrix is ​​recorded as Tuser.

[0062] S5, calculating a weight matrix for each posting information of the user according to the interaction information;

[0063] Each user can obtain a D*1 dimensional weight matrix, where D represents all the user's posting information. The D*1 dimensional weight matrix is ​​recorded as H user The weight matrix calculated based on the interactive information is the "heat" of the post information. The value of the weight matrix can be used as the weight value of the corresponding post information. The weight value indicates the importance of the post information to the user. The higher the value, the more important the topic distribution of the post information is and the stronger its influence is.

[0064] S6, performing weighted processing on the plurality of first topic distribution matrices according to the weight matrix to obtain a plurality of second topic distribution matrices;

[0065] In S5, the weight value indicates the importance of the post information to the user, so the weight matrix H calculated by the interactive information user By weighting the first topic distribution matrix of the post information Tuser, the user's second topic distribution matrix X can be obtained. userSince the first topic distribution matrix is ​​a multidimensional matrix obtained by calculating all the user's posting information using a short text topic model based on semantics and word expansion, each interactive information is weighted to add weight to the corresponding posting information, thereby calculating the user's topic distribution.

[0066] S7, performing vector merging and normalization processing on the multiple second topic distribution matrices to obtain the topic distribution of the posting information of the user.

[0067] For multiple second topic distribution matrices X user , it is necessary to merge all the vectors in multiple matrices to represent the user's topic distribution. However, due to the different number of interactions among users, there are large differences in the values ​​of topic distribution among users. Therefore, normalization processing is required to reduce the differences in topic distribution.

[0068] In another embodiment of the first embodiment of the present application, the preprocessing operation includes performing a word segmentation operation on the post information, removing stop words from the post information, and removing interference symbols from the post information.

[0069] It should be noted that the preprocessing operation may also include part-of-speech tagging of the post information, keyword extraction, topic extraction, summary, event extraction and semantic extraction of the post information, so that the converted text data is more diverse.

[0070] In another embodiment of the first embodiment of the present application, Figure 5 As shown, the steps of constructing a short text topic model based on semantics and word expansion according to the text data are as follows:

[0071] Extracting a first word pair with semantic association in the context of the text data by using a semantic dependency analysis method;

[0072] The keywords to be expanded in the text data are obtained, the keywords to be expanded are input into an external corpus, and the correlation between the keywords to be expanded and words in the external corpus is calculated using point mutual information.

[0073] Specifically, the short text topic model based on semantics and word expansion improves the traditional BTM model. Figure 2 As shown in the figure, the traditional BTM model uses a sliding window approach to obtain word pair information, ignoring the semantic association of the word pair in the context of the text. Figure 3As shown in the figure, the improved BTM model uses the semantic dependency method to extract word pairs in the text, so that the BTM model integrates text semantic information. For example, in the traditional BTM model, taking "xx won the xx country election and became the next president of xx country" as an example, after preprocessing the post information, in the process of constructing the word pair data set, the sliding window will construct "election" and "become" as a word pair. However, based on the understanding and analysis of people's understanding of this sentence, there is no grammatical and logical connection between "election" and "become". It is unreasonable to form a word pair with the two words and enhance the probability that the two words belong to the same topic. There is an obvious semantic and logical connection between the two words "xx" and "president". It is more reasonable to enhance the probability that they belong to the same topic. However, due to the size of the sliding window, "xx" and "president" will not form a word pair. In the first word pair extracted by the semantic dependency method, still taking the sentence "xx won the xx country election and became the next president of xx country" as an example, "xx" and "become" belong to the subject-predicate relationship, "become" and "president" belong to the verb-object relationship, and "xx" and "president" belong to the subject-object relationship. Therefore, in the construction of word pairs, words with strong association relationships are selected to form the first word pair, so that the composed first word pair has the semantic association of the context in the post information.

[0074] At the same time, in order to ensure that the model can further solve the sparse problem when the post information is short text, a word pair expansion method based on word association is proposed. This method is to calculate the correlation between the keyword to be expanded x and the word y in the external corpus through the point mutual information PMI by inputting the keyword to be expanded x into the external corpus. The point mutual information of x and y is recorded as PMI(x,y), as shown in the following formula:

[0075]

[0076] Among them, P(x) is the probability of keyword x appearing in the external corpus; P(y) is the probability of word y appearing in the external corpus. For example, take the words "Xiaomi", "Apple", "Huawei", "Computer" as examples (assuming that these words are extracted from user posts in the text), and extract the target words with the highest PMI value with these words in the news external data set:

[0077] Table 1 Example of word expansion calculation results

[0078]

[0079]

[0080] Finally, the word "Xiaomi" can be expanded into "Xiaomi Coolpad", "Xiaomi MIUI", "Xiaomi Ecosystem", etc. The word "Apple" can be expanded into "Apple Apple", "Apple Jobs", "Apple Cook", etc. The other words are also expanded in a similar way. This solves the sparse problem when the posted information is short text information.

[0081] In another embodiment of the first embodiment of the present application, a judgment threshold of the correlation is set, and if the obtained correlation is greater than the judgment threshold, the keyword to be expanded and the word in the external corpus are combined into a second word pair.

[0082] Specifically, if the correlation value obtained by the correlation calculation satisfies the set judgment threshold, which is generally 200, it is considered that the keyword to be expanded has a strong correlation with the word in the external corpus and can form a second word pair.

[0083] In another embodiment of the first embodiment of the present application, all vectors in the multiple second topic distribution matrices are merged to obtain the user's topic distribution feature vector, and the topic distribution feature vector is normalized using a softmax function to obtain the topic distribution of the user's posting information.

[0084] Specifically, Figure 4 As shown, the existing technical solution combines all posts of the user into a long document and directly inputs it into the traditional BTM model for topic analysis. However, the present invention performs topic analysis on each post of the user separately, and then merges and normalizes the topic distribution of all posts of the same user to use this as the topic feature of the user. The normalization process is shown in the following formula: U topic represents the topic distribution of user posts, X d represents the topic distribution feature vector, and D represents all the user's posting information (how many articles the user has posted).

[0085] Based on the above technical solutions, the social user topic analysis method of Example 1 of the present application proposes a short text topic model based on semantics and word expansion on the basis of the traditional short text topic model. The model performs topic analysis on each social post information to obtain a topic distribution matrix, and then calculates the weight matrix of the post information, that is, the importance of the post information, according to the interactive information of each post information. Finally, the topic distribution matrix and the weight matrix are used to calculate the user's topic distribution. Compared with the user topic analysis method of directly constructing pseudo-documents in the prior art, the social user topic analysis method based on semantics and word expansion of the present invention is more advantageous.

[0086] Embodiment 2:

[0087] Based on the same concept, Figure 6 As shown, this second embodiment provides a social user topic analysis system based on semantics and word expansion on the basis of the first embodiment, including:

[0088] The information acquisition unit 110 is used to acquire the user's posting information and the interactive information generated by the posting information in the social network platform;

[0089] The preprocessing unit 120 is used to perform a preprocessing operation on the posting information to obtain text data of the posting information;

[0090] A model building unit 130, configured to build a short text topic model based on semantics and word expansion according to the text data;

[0091] A first calculation unit 140 is used to input the post information into the short text topic model for calculation to obtain a plurality of first topic distribution matrices of the post information;

[0092] A second calculation unit 150 is used to calculate a weight matrix of each posting information of the user according to the interaction information;

[0093] A weighted processing unit 160, configured to perform weighted processing on the plurality of first topic distribution matrices according to the weight matrix to obtain a plurality of second topic distribution matrices;

[0094] The merging processing unit 170 is used to perform vector merging and normalization processing on the multiple second topic distribution matrices to obtain the topic distribution of the user's posting information.

[0095] The social user topic analysis system of the second embodiment of the present application proposes a short text topic model based on semantics and word expansion on the basis of the traditional short text topic model. The model performs topic analysis on each social post information to obtain a topic distribution matrix, and then calculates the weight matrix of the post information, that is, the importance of the post information, according to the interactive information of each post information. Finally, the topic distribution matrix and the weight matrix are used to calculate the user's topic distribution. Compared with the user topic analysis system of the prior art that directly constructs pseudo-documents, the social user topic analysis system based on semantics and word expansion of the present invention is more advantageous.

[0096] In another embodiment of the second embodiment of the present application, the pre-processing unit 120 includes a word segmentation unit and a removal unit;

[0097] The word segmentation unit is used to perform word segmentation operations on the post information, and the removal unit is used to remove stop words from the post information and remove interference symbols in the post information.

[0098] In another embodiment of the second embodiment of the present application, the model building unit 130 includes a semantic extraction unit and a word expansion unit;

[0099] The semantic extraction unit is used to extract the first word pair with semantic association in the context of the text data by using a semantic dependency analysis method;

[0100] The word expansion unit is used to obtain the keywords to be expanded from the text data, input the keywords to be expanded into an external corpus, and calculate the correlation between the keywords to be expanded and the words in the external corpus using point mutual information.

[0101] In another embodiment of the second embodiment of the present application, the word expansion unit further includes a judgment unit;

[0102] The judgment unit is used to set a judgment threshold of the correlation. If the obtained correlation is greater than the judgment threshold, the keyword to be expanded and the word in the external corpus are combined into a second word pair.

[0103] In yet another embodiment of the second embodiment of the present application, the merging processing unit 170 further includes a vector merging unit and a processing unit;

[0104] The vector merging unit is used to merge all the vectors in the plurality of second topic distribution matrices to obtain a topic distribution feature vector of the user;

[0105] The processing unit is used to normalize the topic distribution feature vector using a softmax function to obtain the topic distribution of the user's posting information.

[0106] It should be noted that the methods executed by the above-mentioned program units in the second embodiment can refer to the various embodiments of the social user topic analysis method of the present invention, and therefore will not be described again.

[0107] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A social user topic analysis method based on semantics and word expansion, characterized in that: The following steps are involved: Obtaining the user's posting information and the interactive information generated by the posting information in the social networking platform; Performing a preprocessing operation on the posting information to obtain text data of the posting information; A short text topic model based on semantics and word expansion is constructed according to the text data; wherein the steps of constructing the short text topic model based on semantics and word expansion are as follows: a semantic dependency analysis method is used to extract a first word pair with semantic association in the context of the text data; a keyword to be expanded of the text data is obtained, the keyword to be expanded is input into an external corpus, and the correlation between the keyword to be expanded and the word in the external corpus is calculated using point mutual information; Inputting the posting information into the short text topic model for calculation to obtain a plurality of first topic distribution matrices of the posting information; Calculate a weight matrix for each posting information of the user according to the interactive information; Performing weighted processing on the plurality of first topic distribution matrices according to the weight matrix to obtain a plurality of second topic distribution matrices; Vector merging and normalization processing are performed on the multiple second topic distribution matrices to obtain the topic distribution of the posting information of the user.

2. A social user topic analysis method based on semantics and word expansion according to claim 1, characterized in that: The preprocessing operation includes performing a word segmentation operation on the post information, removing stop words from the post information, and removing interference symbols from the post information.

3. The social user topic analysis method based on semantics and word expansion according to claim 1 is characterized in that: A judgment threshold of the correlation is set, and if the obtained correlation is greater than the judgment threshold, the keyword to be expanded and the word in the external corpus are combined into a second word pair.

4. The social user topic analysis method based on semantics and word expansion according to claim 1 is characterized in that: All vectors in the multiple second topic distribution matrices are merged to obtain the user's topic distribution feature vector, and the topic distribution feature vector is normalized using a softmax function to obtain the topic distribution of the user's posting information.

5. A social user topic analysis system based on semantics and word expansion, characterized in that: include: An information acquisition unit, used to acquire the user's posting information and the interactive information generated by the posting information in the social network platform; A preprocessing unit, used to perform a preprocessing operation on the posting information to obtain text data of the posting information; A model building unit is used to build a short text topic model based on semantics and word expansion according to the text data; wherein the model building unit includes a semantic extraction unit and a word expansion unit; the semantic extraction unit is used to extract a first word pair with semantic association in the context of the text data by using a semantic dependency analysis method; the word expansion unit is used to obtain a keyword to be expanded from the text data, input the keyword to be expanded into an external corpus, and calculate the correlation between the keyword to be expanded and the word in the external corpus by using point mutual information; A first calculation unit, configured to input the post information into the short text topic model for calculation to obtain a plurality of first topic distribution matrices of the post information; A second calculation unit, used for calculating a weight matrix of each posting information of the user according to the interaction information; A weighted processing unit, used for performing weighted processing on the plurality of first topic distribution matrices according to the weight matrix to obtain a plurality of second topic distribution matrices; The merging processing unit is used to perform vector merging and normalization processing on the multiple second topic distribution matrices to obtain the topic distribution of the user's posting information.

6. A social user topic analysis system based on semantics and word expansion according to claim 5, characterized in that: The preprocessing unit includes a word segmentation unit and a removal unit; The word segmentation unit is used to perform word segmentation operations on the post information, and the removal unit is used to remove stop words from the post information and remove interference symbols in the post information.

7. A social user topic analysis system based on semantics and word expansion according to claim 5, characterized in that: The word expansion unit also includes a judgment unit; The judgment unit is used to set a judgment threshold of the correlation. If the obtained correlation is greater than the judgment threshold, the keyword to be expanded and the word in the external corpus are combined into a second word pair.

8. The social user topic analysis system based on semantics and word expansion according to claim 5, characterized in that: The merging processing unit also includes a vector merging unit and a processing unit; The vector merging unit is used to merge all the vectors in the plurality of the second topic distribution matrices to obtain a topic distribution feature vector of the user; The processing unit is used to normalize the topic distribution feature vector using a softmax function to obtain the topic distribution of the user's posting information.

Images