An Optimization Method for Text Clustering Algorithms for Open Source Intelligence

Abstract

This invention discloses an optimization method for text clustering algorithms targeting open-source intelligence, belonging to the field of artificial intelligence technology. The method includes the following steps: acquiring open-source intelligence text, and performing word segmentation, cleaning, and filtering on the open-source intelligence text to obtain an intelligence vocabulary; calculating the weight value of each keyword in the intelligence vocabulary based on a historical keyword database and a keyword weight model, and selecting the top keywords in the vocabulary from high to low weight values. N A target keyword table is constructed using a set of keywords. The keywords in the target keyword table are vectorized using a bag-of-words model, and several target word vectors are generated based on a TF-IDF model and keyword weight correction coefficients. The target word vectors are then clustered several times using a clustering algorithm to obtain several clustering results. Based on each clustering result, the historical keyword database is updated, thus completing the optimization of the text clustering algorithm for open-source intelligence. This invention solves the problems of insufficient clustering efficiency and accuracy in text clustering algorithms for open-source intelligence.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence technology, and in particular relates to a text clustering algorithm optimization method for open source intelligence. Background Art

[0002] With the development of artificial intelligence, the ability to collect open source intelligence has been greatly improved. Intelligence is the basis of decision-making, but the more the better. It cannot be used without analysis. Valuable intelligence must be extracted through analysis. Clustering analysis of open source intelligence texts through clustering algorithms can effectively organize the collected texts and automatically classify texts with high similarity without the need for prior training.

[0003] Text clustering is an unsupervised machine learning method that does not require a training process or manual pre-labeling of documents. It has certain flexibility and high automation capabilities. Existing clustering algorithms include partition-based clustering algorithms (such as the K-means algorithm), density-based clustering algorithms (such as the DBSCAN algorithm), hierarchical clustering algorithms (such as the HC algorithm), and model-based clustering algorithms (such as the HMM algorithm). Among them, partition-based clustering algorithms and density-based clustering algorithms are the most widely used. Clustering algorithms are calculated for numbers and require text clustering to convert the vocabulary of the text into digital vectors. Commonly used algorithms include the bag-of-words model and word2vec. When clustering algorithms are currently used to perform clustering analysis on open source intelligence, there are still problems such as low efficiency in word segmentation and word selection, and difficulty in accurately meeting the special semantic requirements of special intelligence. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present invention provides a text clustering algorithm optimization method for open source intelligence. Based on the historical keyword library, it updates and records the first several important clustering keywords in each qualified clustering result, and reversely optimizes and extracts the target keyword table of new open source intelligence information clustering tasks in turn. It also improves the calculation effect of the TF-IDF algorithm, ensures that the keywords in the text are converted into effective data vectors, provides a basis for meeting the special semantic requirements of special intelligence, and solves the problem of insufficient clustering efficiency and accuracy of the text clustering algorithm for open source intelligence.

[0005] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0006] The present invention provides a text clustering algorithm optimization method for open source intelligence, comprising the following steps:

[0007] S1. Obtain open source intelligence text, and perform word segmentation, cleaning and filtering on the open source intelligence text to obtain an intelligence word list;

[0008] S2. According to the historical keyword database, the weight value of each keyword in the intelligence vocabulary is calculated based on the keyword weight model, and the first N keywords ranked from high to low in weight value are selected to construct the target keyword table;

[0009] S3, using the bag-of-words model to vectorize the keywords in the target keyword table, and generating several target word vectors based on the TF-IDF model and the keyword weight correction coefficient;

[0010] S4. According to the clustering algorithm, cluster the target word vector several times to obtain several clustering results;

[0011] S5. Based on the clustering results, optimize and update the historical keyword library to complete the text clustering algorithm optimization for open source intelligence.

[0012] The beneficial effects of the present invention are as follows: the present invention provides a text clustering algorithm optimization method for open source intelligence, which effectively extracts words with intelligence significance and forms an intelligence vocabulary by segmenting, cleaning and filtering the open source intelligence text; based on the historical keyword library, the present invention calculates the weight value of each keyword in the intelligence vocabulary based on the keyword weight model, providing a basis for selecting keywords with high weight values ​​for unsupervised clustering, and the keywords with high weight values ​​reflect the open source intelligence information with high attention in the historical keyword library, which can effectively improve the text clustering effect of the open source intelligence information; after the present invention uses the bag-of-words model to vectorize the keywords in the target keyword table, based on The TF-IDF model corrects the TF-IDF weight of the word vector in combination with the keyword weight correction coefficient, generates a number of target word vectors, and optimizes the clustering results through the corrected TF-IDF statistical indicators, thereby significantly improving the clustering result quality of the open source intelligence text; the present invention clusters the target word vectors through an unsupervised clustering algorithm, and optimizes and updates the historical keyword library based on the clustering results, thereby achieving the situation where the requirements of the open source intelligence text information acquisition task are met, by cyclically updating the historical keyword library, and continuously cyclically enhancing the sensitivity to the keywords corresponding to the target intelligence, thereby significantly improving the clustering result quality of the open source intelligence text.

[0013] Furthermore, the S1 comprises the following steps:

[0014] S11. Obtain open source intelligence texts;

[0015] S12. Use NLP tools to perform word segmentation and part-of-speech tagging on the open-source intelligence text to obtain the text after word segmentation and tagging;

[0016] S13, cleaning and deleting punctuation marks, special characters and tags in the text after the word segmentation annotation, and filtering and deleting stop words in the text after the word segmentation annotation according to a stop word list, to obtain the remaining words after cleaning and filtering;

[0017] S14, performing a secondary filtering on the remaining words after cleaning and filtering according to parts of speech to obtain nouns and verbs in the text;

[0018] S15. Use nouns and verbs in the text as keywords and construct an intelligence vocabulary.

[0019] The beneficial effect of adopting the above further scheme is: the present invention provides a method for cleaning and deleting factors without intelligence significance in open source intelligence texts, and filters and retains nouns and verbs in articles with intelligence significance. The intelligence vocabulary constructed based on the vocabulary with intelligence significance can provide a basis for constructing a target keyword table and selecting target keywords to improve the clustering quality of open source intelligence texts.

[0020] Furthermore, the calculation expression of the keyword weight model in S2 is as follows:

[0021] ,

[0022] in, represents the weight value of the keyword, x represents the score of the keyword in the historical keyword library, y represents the number of times the keyword appears in the open source intelligence text, and x i represents the score of the i-th keyword in the intelligence vocabulary in the historical keyword library, y i It represents the number of times the i-th keyword in the intelligence vocabulary appears in the open source intelligence text, i represents the i-th keyword in the intelligence vocabulary, and n represents the total number of keywords in the intelligence vocabulary.

[0023] The beneficial effect of adopting the above further scheme is: the present invention provides a method for calculating the keyword weight model, combining the corresponding scores of the keywords in the intelligence vocabulary in the historical keyword library, the total number of occurrences of all keywords in the open source intelligence text and the number of occurrences of each keyword in the open source intelligence text, and calculates the keyword weight. The keyword weight reflects the closeness between the keyword and the target information to be obtained by the open source intelligence. Keywords based on high weight values ​​can effectively optimize and improve the quality of clustering results of open source intelligence.

[0024] Furthermore, S3 comprises the following steps:

[0025] S31, using the bag-of-words model to vectorize the keywords in the target keyword table to obtain a number of word vectors, wherein the dimension of the word vector is consistent with the number of keywords in the target keyword table, which is N;

[0026] S32, calculate the word frequency and reverse file frequency corresponding to the word vector;

[0027] S33, multiplying the word frequency corresponding to the word vector and the inverse document frequency to obtain the TF-IDF weight corresponding to the word vector;

[0028] S34, setting keyword weight correction coefficient;

[0029] S35, based on the keyword weight correction coefficient, the TF-IDF weight corresponding to each word vector is corrected to obtain a corrected TF-IDF weight corresponding to each word vector;

[0030] S36. Generate several target word vectors based on the TF-IDF model according to the modified TF-IDF weights corresponding to each word vector.

[0031] The beneficial effect of adopting the above further scheme is: the present invention provides a method for correcting the TF-IDF statistical indicators of word vectors after keyword vectorization processing, combining the TF-IDF model and the keyword weight correction coefficient. The target word vector generated after the TF-IDF statistical indicator correction enhances the category distinction ability of keywords, effectively improves the correlation sensitivity between keywords and target acquired open source intelligence information, and provides a basis for improving the text clustering ability and quality of open source intelligence.

[0032] Furthermore, the calculation expressions of the word frequency and the reverse file frequency corresponding to the word vector in S32 are as follows:

[0033] ,

[0034] ,

[0035] in, Indicates The keyword corresponding to the word vector is in the first The normalized word frequency in the article, Indicates The keyword corresponding to the word vector is in the first The number of times it appears in an article, k represents the total number of word vectors, Indicates the first The total number of times the keywords corresponding to all word vectors in the article appear, Indicates The inverse document frequency corresponding to the word vector, represents the total number of articles in the open source intelligence text, lg represents the logarithmic operation, Indicates that it contains The total number of articles with keywords corresponding to the word vectors, Indicates that it contains The article with the keyword corresponding to the word vector, Indicates the first Articles.

[0036] The beneficial effect of adopting the above further scheme is: the present invention provides a method for calculating the word frequency and reverse file frequency corresponding to the word vector. The importance of a keyword increases in direct proportion to the number of times it appears in the text, but at the same time decreases in inverse proportion to the frequency of its appearance in the corpus. The word frequency and reverse file frequency corresponding to the word vector can effectively reflect the degree of correlation between the open source intelligence text and the target intelligence information content to a certain extent. On this basis, it also provides an improvement basis for the scheme of the present invention to further enhance the category distinction ability of keywords in combination with the keyword weight correction coefficient to meet special semantic requirements.

[0037] Furthermore, the calculation expression of the keyword weight correction coefficient in S34 is as follows:

[0038] ,

[0039] in, Indicates The keyword weight correction coefficient corresponding to the word vector, Indicates The frequency of the keyword corresponding to the word vector in the historical keyword library, Indicates that it contains The first keyword corresponding to the word vector The total number of keywords in the article. Indicates The sum of the word frequencies of the keywords corresponding to the word vectors in the historical keyword library.

[0040] The beneficial effect of adopting the above further scheme is: the present invention provides a method for calculating the keyword weight correction coefficient, which fully considers the word frequency of the keyword corresponding to the word vector in the historical keyword library, the total number of all keywords in the open source intelligence articles where the keyword appears, and the sum of the word frequencies of the keywords corresponding to all word vectors in the historical keyword library, which can effectively improve the accuracy of the TF-IDF weight of the keyword corresponding to the word vector in the open source intelligence text article, thereby significantly improving the accuracy of the open source intelligence text division and clustering.

[0041] Furthermore, the calculation expression of the modified TF-IDF weight corresponding to the word vector in S35 is as follows:

[0042] ,

[0043] in, Indicates The keyword corresponding to the word vector is in the first The modified TF-IDF weights in the article, Indicates The keyword corresponding to the word vector is in the first TF-IDF weight in the article.

[0044] The beneficial effect of adopting the above further scheme is: the present invention provides a method for calculating the modified TF-IDF weight corresponding to the word vector, and realizes the correction of the TF-IDF statistical index of the word vector by combining the keyword weight correction coefficient with the TF-IDF model, thereby enhancing the classification ability of the target word vector and providing a basis for improving the clustering efficiency and accuracy of open source intelligence.

[0045] Furthermore, the S5 comprises the following steps:

[0046] S51, according to the clustering target, selecting qualified clustering results to be confirmed from the clustering results as target clustering results;

[0047] S52, according to the target clustering result, obtain the keywords corresponding to the first M largest target word vectors from the center point of each cluster, and form a keyword table in the form of a two-dimensional array as the basis for the current clustering;

[0048] S53: Update the keywords corresponding to the acquired target word vector into the historical keyword library, and update the scores corresponding to the keywords in the historical keyword library accordingly, so as to complete the text clustering algorithm optimization for open source intelligence.

[0049] The beneficial effect of adopting the above further scheme is: the present invention provides a method for optimizing and updating the historical keyword library based on each clustering result, by selecting the target clustering result from the qualified clustering results, and selecting the keyword corresponding to the target word vector from the center point of the cluster cluster in the target clustering result to update it to the historical keyword library, and synchronously updating the score corresponding to each keyword in the historical keyword library, which can realize the cyclic optimization of the historical keyword library, thereby continuously improving the correlation between the historical keyword library and the target acquired open source intelligence information, as well as the text clustering efficiency and accuracy for open source intelligence.

[0050] Furthermore, when the keywords corresponding to the acquired target word vectors are updated into the historical keyword library, the qualified time of the clustering results and the corresponding clustering algorithm are recorded at the same time.

[0051] The beneficial effect of adopting the above further scheme is: the present invention provides an analytical basis for later analyzing the efficiency and accuracy of open source intelligence clustering from different dimensions such as time and clustering method by recording the qualified selection time of the clustering results and the corresponding clustering algorithm adopted, and provides a more dimensional open source intelligence clustering optimization basis and optimization direction.

[0052] Other advantages of the present invention will be analyzed in more detail in subsequent embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] 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.

[0054] Figure 1 The present invention is a flowchart of the steps of a text clustering algorithm optimization method for open source intelligence in an embodiment of the present invention. DETAILED DESCRIPTION

[0055] The technical solutions 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 a part of the embodiments of the present invention, rather than all the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present invention.

[0056] like Figure 1 As shown, in one embodiment of the present invention, the present invention provides a text clustering algorithm optimization method for open source intelligence, comprising the following steps:

[0057] S1. Obtain open source intelligence text, and perform word segmentation, cleaning and filtering on the open source intelligence text to obtain an intelligence word list;

[0058] The S1 comprises the following steps:

[0059] S11. Obtain open source intelligence texts;

[0060] S12. Use NLP tools to perform word segmentation and part-of-speech tagging on the open-source intelligence text to obtain the text after word segmentation and tagging;

[0061] S13, cleaning and deleting punctuation marks, special characters and tags in the text after the word segmentation annotation, and filtering and deleting stop words in the text after the word segmentation annotation according to a stop word list, to obtain the remaining words after cleaning and filtering;

[0062] In this embodiment, special characters include arrow symbols, sticker characters, Russian uppercase and lowercase characters, phonetic symbols, pinyin, Japanese hiragana, numerical serial numbers, unit symbols, mathematical symbols, etc.; the stop word list includes 1893 common Chinese stop words such as !, #, $, %, &, etc.

[0063] S14, performing a secondary filtering on the remaining words after cleaning and filtering according to parts of speech to obtain nouns and verbs in the text;

[0064] S15. Use nouns and verbs in the text as keywords and construct an intelligence vocabulary.

[0065] S2. According to the historical keyword database, the weight value of each keyword in the intelligence vocabulary is calculated based on the keyword weight model, and the first N keywords ranked from high to low in weight value are selected to construct the target keyword table;

[0066] The calculation expression of the keyword weight model in S2 is as follows:

[0067] ,

[0068] in, represents the weight value of the keyword, x represents the score of the keyword in the historical keyword library, y represents the number of times the keyword appears in the open source intelligence text, and x i represents the score of the i-th keyword in the intelligence vocabulary in the historical keyword library, y i It represents the number of times the i-th keyword in the intelligence vocabulary appears in the open source intelligence text, i represents the i-th keyword in the intelligence vocabulary, and n represents the total number of keywords in the intelligence vocabulary.

[0069] S3, using the bag-of-words model to vectorize the keywords in the target keyword table, and generating several target word vectors based on the TF-IDF model and the keyword weight correction coefficient;

[0070] The S3 comprises the following steps:

[0071] S31, using the bag-of-words model to vectorize the keywords in the target keyword table to obtain a number of word vectors, wherein the dimension of the word vector is consistent with the number of keywords in the target keyword table, which is N;

[0072] S32, calculate the word frequency and reverse file frequency corresponding to the word vector;

[0073] The calculation expressions of the word frequency and the reverse file frequency corresponding to the word vector in S32 are as follows:

[0074] ,

[0075] ,

[0076] in, Indicates The keyword corresponding to the word vector is in the first The normalized word frequency in the article, Indicates The keyword corresponding to the word vector is in the first The number of times it appears in an article, k represents the total number of word vectors, Indicates the first The total number of times the keywords corresponding to all word vectors in the article appear, Indicates The inverse document frequency corresponding to the word vector, represents the total number of articles in the open source intelligence text, lg represents the logarithmic operation, Indicates that it contains The total number of articles with keywords corresponding to the word vectors, Indicates that it contains The article with the keyword corresponding to the word vector, Indicates the first In this embodiment, the word frequency refers to the number of times the keyword appears in the article.

[0077] S33, multiplying the word frequency corresponding to the word vector and the inverse document frequency to obtain the TF-IDF weight corresponding to the word vector;

[0078] S34, setting keyword weight correction coefficient;

[0079] The calculation expression of the keyword weight correction coefficient in S34 is as follows:

[0080] ,

[0081] in, Indicates The keyword weight correction coefficient corresponding to the word vector, Indicates The frequency of the keyword corresponding to the word vector in the historical keyword library, Indicates that it contains The first word of the keyword corresponding to the word vector The total number of keywords in the article. Indicates The sum of the word frequencies of the keywords corresponding to the word vectors in the historical keyword library.

[0082] S35, based on the keyword weight correction coefficient, the TF-IDF weight corresponding to each word vector is corrected to obtain a corrected TF-IDF weight corresponding to each word vector;

[0083] The calculation expression of the modified TF-IDF weight corresponding to the word vector in S35 is as follows:

[0084] ,

[0085] in, Indicates The keyword corresponding to the word vector is in the first The modified TF-IDF weights in the article, Indicates The keyword corresponding to the word vector is in the first TF-IDF weight in the article.

[0086] S36. Generate several target word vectors based on the TF-IDF model according to the modified TF-IDF weights corresponding to each word vector.

[0087] S4. According to the clustering algorithm, cluster the target word vector several times to obtain several clustering results;

[0088] The clustering algorithm used in this embodiment can be any common clustering algorithm, such as any clustering algorithm based on partitioning, density-based clustering algorithm, etc.

[0089] S5. Based on the clustering results, optimize and update the historical keyword library to complete the text clustering algorithm optimization for open source intelligence.

[0090] The S5 comprises the following steps:

[0091] S51, according to the clustering target, selecting qualified clustering results to be confirmed from the clustering results as target clustering results;

[0092] In this embodiment, the qualified clustering results to be confirmed are clustering results to be confirmed that meet the requirements of the open source intelligence information acquisition task.

[0093] S52. According to the target clustering result, the keywords corresponding to the first M largest target word vectors are obtained from the center point of each clustering cluster, and a keyword table in the form of a two-dimensional array is constructed as the basis for the current clustering; in this embodiment, the keywords corresponding to the first M largest target word vectors are: the keywords corresponding to the first M word vectors after the word vectors are sorted from large to small.

[0094] In this embodiment, the keyword table in the form of a two-dimensional array is shown in Table 1:

[0095] Table 1

[0096]

[0097] S53: Update the keywords corresponding to the acquired target word vector into the historical keyword library, and update the scores corresponding to the keywords in the historical keyword library accordingly, so as to complete the text clustering algorithm optimization for open source intelligence.

[0098] When the keywords corresponding to the acquired target word vectors are updated to the historical keyword library, the qualified time of the clustering results and the corresponding clustering algorithm are recorded at the same time.

[0099] In this embodiment, when the successful frequency of clustering statistics is calculated, corresponding analysis can be performed according to the recording conditions. All clustering times can be counted, or conditional statistics can be performed based on the selected qualified time and the clustering algorithm used, thereby satisfying clustering optimization analysis in different dimensions.

[0100] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A text clustering algorithm optimization method for open source intelligence, It is characterized in that The steps include: S1. Obtain open source intelligence text, and perform word segmentation, cleaning and filtering on the open source intelligence text to obtain an intelligence word list; S2. According to the historical keyword database, the weight value of each keyword in the intelligence vocabulary is calculated based on the keyword weight model, and the top keywords ranked from high to low are selected. N Build the target keyword list based on keywords; The calculation expression of the keyword weight model in S2 is as follows: ， in, Indicates the weight value of the keyword. x Indicates the score of the keyword in the historical keyword database. y Indicates the number of times a keyword appears in the open source intelligence text. x i Indicates the first i The score corresponding to a keyword in the historical keyword database, y i Indicates the first i The number of times a keyword appears in the open source intelligence text, i Indicates the first i Keywords, n Indicates the total number of keywords in the intelligence vocabulary; S3, using the bag-of-words model to vectorize the keywords in the target keyword table, and generating several target word vectors based on the TF-IDF model and the keyword weight correction coefficient; S4. According to the clustering algorithm, the target word vector is clustered several times to obtain several clustering results; S5. Based on the clustering results, optimize and update the historical keyword library to complete the text clustering algorithm optimization for open source intelligence; The S5 comprises the following steps: S51, according to the clustering target, selecting qualified clustering results to be confirmed from the clustering results as target clustering results; S52, according to the target clustering result, obtain the previous M The keywords corresponding to the largest target word vectors are selected and a keyword table in the form of a two-dimensional array is formed as the basis for the current clustering; S53: Update the keywords corresponding to the acquired target word vector into the historical keyword library, and update the scores corresponding to the keywords in the historical keyword library accordingly, so as to complete the optimization of the text clustering algorithm for open source intelligence.

2. According to the text clustering algorithm optimization method for open source intelligence according to claim 1, It is characterized in that The S1 comprises the following steps: S11. Obtain open source intelligence texts; S12. Use NLP tools to perform word segmentation and part-of-speech tagging on the open-source intelligence text to obtain the text after word segmentation and tagging; S13, cleaning and deleting punctuation marks, special characters and tags in the text after the word segmentation annotation, and filtering and deleting stop words in the text after the word segmentation annotation according to a stop word list, to obtain the remaining words after cleaning and filtering; S14, performing a secondary filtering on the remaining words after cleaning and filtering according to parts of speech to obtain nouns and verbs in the text; S15. Use nouns and verbs in the text as keywords and construct an intelligence vocabulary.

3. According to the text clustering algorithm optimization method for open source intelligence according to claim 1, It is characterized in that The S3 comprises the following steps: S31. Use the bag-of-words model to vectorize the keywords in the target keyword table to obtain a number of word vectors, where the dimension of the word vector is consistent with the number of keywords in the target keyword table, which is N ; S32, calculate the word frequency and reverse file frequency corresponding to the word vector; S33, multiplying the word frequency corresponding to the word vector and the inverse document frequency to obtain the TF-IDF weight corresponding to the word vector; S34, setting keyword weight correction coefficient; S35, based on the keyword weight correction coefficient, the TF-IDF weight corresponding to each word vector is corrected to obtain a corrected TF-IDF weight corresponding to each word vector; S36. Generate several target word vectors based on the TF-IDF model according to the modified TF-IDF weights corresponding to each word vector.

4. The text clustering algorithm optimization method for open source intelligence according to claim 3, It is characterized in that The calculation expressions of the word frequency and the reverse file frequency corresponding to the word vector in S32 are as follows: ， ， in, Indicates The keyword corresponding to the word vector is in the first The normalized word frequency in the article, Indicates The keyword corresponding to the word vector is in the first The number of times it appears in an article, k represents the total number of word vectors, Indicates the first The total number of times the keywords corresponding to all word vectors in the article appear, Indicates The inverse document frequency corresponding to the word vector, represents the total number of articles in the open source intelligence text, lg represents the logarithmic operation, Indicates that it contains The total number of articles with keywords corresponding to the word vectors, Indicates that it contains The article with the keyword corresponding to the word vector, Indicates the first Articles.

5. The text clustering algorithm optimization method for open source intelligence according to claim 4, It is characterized in that The calculation expression of the keyword weight correction coefficient in S34 is as follows: ， in, Indicates The keyword weight correction coefficient corresponding to the word vector, Indicates The frequency of the keyword corresponding to the word vector in the historical keyword library, Indicates that it contains The first keyword corresponding to the word vector The total number of keywords in the article. Indicates The sum of the word frequencies of the keywords corresponding to the word vectors in the historical keyword library.

6. The text clustering algorithm optimization method for open source intelligence according to claim 5, It is characterized in that The calculation expression of the modified TF-IDF weight corresponding to the word vector in S35 is as follows: ， in, Indicates The keyword corresponding to the word vector is in the first The modified TF-IDF weights in the article, Indicates The keyword corresponding to the word vector is in the first TF-IDF weight in the article.

7. The text clustering algorithm optimization method for open source intelligence according to claim 1, It is characterized in that When the keywords corresponding to the acquired target word vectors are updated to the historical keyword library, the qualified time of the clustering results and the corresponding clustering algorithm are recorded at the same time.

Images