A semantic association retrieval method for local area network digital archives

Abstract

The present invention relates to the field of data processing technology and proposes a semantic association retrieval method for digital archives in a local area network. The method comprises: performing a semantic association retrieval on all text archives in the local area network, collecting a user's search text, and obtaining a word vector for each word in the word segmentation results of the search text; marking a first word and a second word, respectively determining the positional association and correction weight of the first word and the second word, and then determining the semantic weights of the first word and the second word, determining a sentence vector for the search text, and determining a sentence vector for each sentence in all text archives in the local area network; respectively determining a file vector for each text archive in the local area network; and completing a semantic association retrieval of digital archives based on the similarity between the file vectors of all text archives in the local area network and the sentence vector of the search text. The present invention aims to improve the accuracy of semantic association retrieval of digital archives.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a semantic association retrieval method for local area network digital archives. Background Art

[0002] Semantic association retrieval of text files in digital archives involves in-depth understanding and semantic analysis of the text file content to achieve association and retrieval between files. In semantic retrieval, semantic similarity between files is typically calculated to achieve more accurate retrieval. Specifically, file retrieval can be performed based on the semantic similarity between the user's search information and the files.

[0003] However, only using the word vector similarity between the search information entered by the user and the database text cannot combine the contextual information in the file, which often leads to large semantic differences between the file in the matching result and the search information entered by the user, resulting in the inability to retrieve the user's target file. Summary of the Invention

[0004] The present invention provides a semantic association retrieval method for local area network digital archives to solve the problem that digital archive retrieval cannot associate the context information in the archives, resulting in inaccurate retrieval matching results. The technical solution adopted is as follows:

[0005] An embodiment of the present invention provides a semantic association retrieval method for local area network digital archives, the method comprising the following steps:

[0006] Perform semantic association search on all text files in the local area network, collect the user's search text, and obtain the word vector of each word in the word segmentation results of the search text;

[0007] Record any two different words in the word segmentation results of the search text as the first word and the second word respectively; determine the positional relevance of the first word and the second word based on the positional difference between the first word and the second word in the search text; determine the modified weights of the first word and the second word based on the sentence length and number of times the first word and the second word appear simultaneously in the same sentence in all text files in the local area network, as well as the semantic relevance of the first word and the second word; determine the semantic weights of the first word and the second word based on the positional relevance and the modified weight; determine the sentence vector of the search text based on the modified weights between all words in the word segmentation results of the search text and the word vectors of all words; and determine the sentence vector of each sentence in all text files in the local area network;

[0008] Determine a file vector for each text file in the local area network according to the number of times all different words in each sentence in all text files in the local area network co-appear in the same sentence in all text files in the local area network and the sentence vector of each sentence in all text files in the local area network;

[0009] According to the similarity between the file vectors of all text files in the local area network and the sentence vectors of the search text, the semantic association retrieval of digital files is completed.

[0010] Furthermore, the method for determining the positional relevance between the first word and the second word is:

[0011] The absolute value of the difference between the sequence numbers of the first word and the second word is recorded as the position distance between the first word and the second word, and the negative correlation processing result of the position distance between the first word and the second word is recorded as the position relevance between the first word and the second word.

[0012] Furthermore, the method of determining the modified weights of the first word and the second word based on the length and number of times the first word and the second word appear simultaneously in the same sentence in all text files in the local area network, and the semantic relevance between the first word and the second word, includes the following specific methods:

[0013] Use the TextRank algorithm to obtain the weight of the edge between the first word and the second word;

[0014] When a first word and a second word appear simultaneously in the same sentence in all text files in the local area network, the first word and the second word are recorded as co-occurring in the text files; a first ratio of the first word to the second word is determined based on the difference between the number of co-occurrences of the first word and the second word and the number of co-occurrences of all different words in all text files in the local area network;

[0015] determining a second ratio of the first word to the second word based on the lengths of all sentences in all text files in the local area network and the lengths of sentences in which the first word and the second word co-occur in the text files;

[0016] The average of the first ratio and the second ratio of the first word to the second word is recorded as the third ratio of the first word to the second word;

[0017] The difference between the weight of the edge between the first word and the second word and the third ratio is recorded as the modified weight of the first word and the second word.

[0018] Furthermore, the method of determining the first ratio of the first word to the second word based on the difference between the number of co-occurrences of the first word and the second word and the number of co-occurrences of all different words in all text files in the local area network includes the following specific methods:

[0019] The total number of times the first word and the second word co-occur in all text files in the local area network is recorded as the document co-occurrence number of the first word and the second word, the total number of times all different words co-occur in all text files in the local area network is recorded as the total document co-occurrence number, and the ratio of the document co-occurrence number of the first word and the second word to the total document co-occurrence number is recorded as the first ratio of the first word and the second word.

[0020] Furthermore, the lengths of all sentences in all text files in the local area network, as well as the lengths of sentences in which the first word and the second word co-occur in the text files, are used to determine a second ratio of the first word to the second word, including the following specific method:

[0021] The average length of the sentences in which the first word and the second word co-occur in the text file is recorded as the document co-occurrence length of the first word and the second word; the average length of all sentences in all text files in the local area network is recorded as the average document length; the ratio of the document co-occurrence length of the first word and the second word to the average document length is recorded as the second ratio of the first word and the second word.

[0022] Furthermore, the semantic weights of the first word and the second word are determined as follows:

[0023] The average of the positional relevance and the modified weight of the first word and the second word is recorded as the semantic weight of the first word and the second word.

[0024] Furthermore, the sentence vector of the search text is determined by:

[0025] The normalized value of the cumulative sum of the modified weights between the first word and all other words in the word segmentation results of the search text is recorded as the total semantic weight of the first word, and the product of the total semantic weight of the first word and the word vector is recorded as the weighted cumulative value of the first word;

[0026] The sum of the weighted cumulative values ​​of all words in the word segmentation results of the search text is recorded as the sentence vector of the search text.

[0027] Furthermore, the file vector of the text file is determined by:

[0028] Record any sentence in all text files in the local area network as a target file sentence, record the cumulative sum of the document co-occurrence times of all different words in the word segmentation results of the target file sentence as the co-occurrence frequency of the target file sentence, record the ratio of the co-occurrence frequency of the target file sentence to the total number of document co-occurrences as the fourth ratio of the target file sentence, and record the product of the sentence vector of the target file sentence and the fourth ratio as the weighted vector of the target file sentence;

[0029] The vector sum of the weighted vectors of all sentences in the same text file in the local area network is recorded as the file vector of the same text file in the local area network.

[0030] Furthermore, the semantic association retrieval of digital archives is completed based on the similarity between the archive vectors of all text archives in the local area network and the sentence vector of the search text, including the specific method of:

[0031] The absolute value of the cosine similarity between the sentence vector of the search text and the file vector of the text file in the local area network is recorded as the semantic correlation degree between the search text and the text file;

[0032] According to the semantic relevance between the search text and all text files in the local area network, the semantic relevance search results of the digital files are obtained.

[0033] Furthermore, the method of obtaining the semantic association search results of digital archives based on the semantic association degree between the search text and all text archives in the local area network includes the following specific methods:

[0034] The largest degree of semantic relevance among all the semantic relevances corresponding to the retrieved text The text files corresponding to the semantic association degrees are arranged in descending order according to the semantic association degrees corresponding to the text files to obtain the semantic association retrieval results, among which, Indicates preset parameters.

[0035] The beneficial effects of the present invention are:

[0036] This application compares the physical distances between different words in a sentence based on the order differences of all words contained in the word segmentation results of the search text, evaluates the degree of intrinsic semantic connection between different words in a sentence, obtains the positional relevance of any two different words in the word segmentation results of the search text, and takes into account that the meanings of the same word may be different in different contexts. In order to avoid the influence of different meanings of the same word on the results of semantic association analysis, the TextRank algorithm adjusts the evaluation of the relevance between words in a sentence, evaluates the semantic relevance of any two different words in the word segmentation results of the search text, obtains the corrected weights of the two different words, and determines the semantic weight in combination with the positional relevance. The semantic weight is a comprehensive relevance evaluation of different words in the word segmentation results of the search text, which can avoid the semantic association caused by single reliance on semantic generalization. The deviation of sex evaluation is eliminated. At the same time, the dependence of traditional retrieval on literal matching is broken through. Further, the sentence vector of the retrieval text is determined according to the corrected weights between all words in the word segmentation results of the retrieval text and the word vectors of all words. In the same way, the sentence vector of each sentence in all text files in the local area network is determined according to all text files in the local area network. On the basis of the sentence vector, the importance of each sentence in all text files in the local area network relative to the text file is determined, and the file vector of each text file in the local area network is determined respectively. Finally, according to the similarity between the file vectors of all text files in the local area network and the sentence vector of the retrieval text, the semantic association retrieval of digital files is completed, which solves the problem that digital file retrieval cannot associate the context information in the file, resulting in inaccurate retrieval matching results, and improves the accuracy of the semantic association retrieval results of digital files in the local area network. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0038] Figure 1 A flowchart of a semantic association retrieval method for digital archives in a local area network provided by one embodiment of the present invention;

[0039] Figure 2 This is a flowchart of obtaining location relevance provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0040] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making creative efforts are within the scope of protection of the present invention.

[0041] See also Figure 1 , which shows a flow chart of a semantic association retrieval method for a local area network digital archive provided by an embodiment of the present invention, the method comprising the following steps:

[0042] Step S001: perform semantic association search on all text files in the local area network, collect the user's search text, and obtain the word vector of each word in the word segmentation result of the search text.

[0043] Digital archives include archives stored in various data formats such as text, images, voice, and video. This embodiment uses text archives as an example to perform semantic association retrieval.

[0044] When performing a semantic association search on text files, the user must first enter the search text. The search text is extracted and segmented using the Jieba word segmentation tool to obtain the segmentation results. Each word in the segmentation results is processed using the Word2vec model to obtain the word vector for each word in the segmentation results.

[0045] Among them, using the jieba word segmentation tool for word segmentation processing and using the Word2vec model to obtain word vectors are both well-known technologies and will not be repeated here. Each word vector in the word segmentation result obtained using the Word2vec model is of equal length. In this embodiment, the length of the word vector is set to 512.

[0046] At this point, the word vector of each word in the word segmentation results of the search text is obtained.

[0047] Step S002: record any two different words in the word segmentation results of the search text as the first word and the second word respectively, determine the positional relevance of the first word and the second word based on the positional difference between the first word and the second word in the search text, determine the correction weights of the first word and the second word based on the sentence length and number of times the first word and the second word appear simultaneously in the same sentence in all text files in the local area network, and the semantic relevance between the first word and the second word, determine the semantic weights of the first word and the second word based on the positional relevance and the correction weight of the first word and the second word, determine the sentence vector of the search text based on the correction weights between all words in the word segmentation results of the search text and the word vectors of all words, and determine the sentence vector of each sentence in all text files in the local area network.

[0048] All words contained in the word segmentation results of the search text are arranged in sequence according to the order of the words in the search text, and the order of the arranged words in all words is recorded as the sequence number of the arranged words.

[0049] For example, when the user's search text is "The cat chases the mouse in the sunshine", the word segmentation results of the search text include the four words "cat", "sunshine", "chase", and "mouse". The sequence numbers of the four words "cat", "sunshine", "chase", and "mouse" are 1, 2, 3, and 4 respectively.

[0050] When performing semantic association retrieval on text archives, semantic association matching is performed on the search text against all archives in the digital archive. Based on the semantic features of the search text, the intrinsic semantic relationships between different words in the search text's segmentation results are extracted. Generally, words that are closer together in the search text have greater intrinsic semantic connections. Therefore, the positional associations of different words in the search text's segmentation results are determined based on the differences in the sequence numbers of different words in the search text's segmentation results.

[0051] Any two different words in the word segmentation results of the search text are respectively recorded as the first word and the second word, the absolute value of the difference between the sequence numbers of the first word and the second word is recorded as the position distance between the first word and the second word, and the negative correlation processing result of the position distance between the first word and the second word is recorded as the position correlation between the first word and the second word.

[0052] It is understandable that the positional distance between the first word and the second word is negatively correlated, that is, it is ensured that the positional distance between the first word and the second word is negatively correlated with the positional relevance between the first word and the second word. It is understandable that the negative correlation in this application refers to the relationship between the independent variable and the dependent variable, the independent variable is the positional distance between the first word and the second word, and the dependent variable is the positional relevance between the first word and the second word. The negative correlation is that the dependent variable decreases (increases) as the independent variable increases (decreases), which can be an inverse relationship, a subtractive relationship, etc.

[0053] Preferably, as an embodiment of the present application, the positional relevance of the first word and the second word is recorded as the positional relevance of the first word and the second word, with the opposite number of the positional distance between the first word and the second word as the exponent and a natural constant as the base.

[0054] In some other embodiments of the present application, the difference between the normalized values ​​of the positional distances between the number 1 and the first word and the second word may be recorded as the positional relevance between the first word and the second word.

[0055] When the distance between the first word and the second word in the word segmentation results of the search text is greater, the possibility that the intrinsic semantic connection between the first word and the second word is smaller is greater. In this case, the position correlation between the first word and the second word is smaller. The position correlation acquisition flow chart is as follows: Figure 2 shown.

[0056] At this point, the positional relevance of any two different words in the word segmentation results of the search text can be obtained.

[0057] Furthermore, the intrinsic correlation between any two different words in the word segmentation results of the search text and the files to be matched is analyzed.

[0058] The meaning of the same word may vary in different contexts. Traditional methods, such as using cosine similarity between word vectors as the similarity measure for semantic association analysis, are susceptible to variations in the meaning of the same word, leading to inaccurate results. The TextRank algorithm models words in a text as nodes in a graph. It constructs edges based on the relationships between nodes, as indicated by word co-occurrence or sentence similarity, assigns weights to these edges, and then calculates the importance score of each node using the relationships between nodes in the graph. Therefore, the TextRank algorithm can be used to determine the relevance between any two distinct words in the segmentation results of a retrieved text. However, since infrequent words often co-occur less frequently in a document, it is easy to overweight infrequent co-occurring words, resulting in inaccurate node importance scores. Therefore, adaptive weighting is necessary in the TextRank algorithm.

[0059] Use the TextRank algorithm to process all text files in the local area network and obtain the weight of the edge between the first word and the second word.

[0060] The use of the TextRank algorithm to obtain the weights of the edges between words is a well-known technique and will not be described in detail.

[0061] When the first and second words appear simultaneously in the same sentence across all text files in the local area network, this is recorded as the co-occurrence of the first and second words in the text file. The total number of times the first and second words co-occur across all text files in the local area network is recorded as the document co-occurrence count of the first and second words. The total number of times all different words co-occur across all text files in the local area network is recorded as the total document co-occurrence count. The ratio of the document co-occurrence count of the first and second words to the total document co-occurrence count is recorded as the first ratio of the first and second words. The average length of the sentences in which the first and second words co-occur in the text file is recorded as the document co-occurrence length of the first and second words. The average length of all sentences in all text files in the local area network is recorded as the average document length. The ratio of the document co-occurrence length of the first and second words to the average document length is recorded as the second ratio of the first and second words. The average of the first and second ratios of the first and second words is recorded as the third ratio of the first and second words. The difference between the weight of the edge between the first and second words and the third ratio is recorded as the modified weight of the first and second words.

[0062] It should be noted that when the first word and the second word do not appear simultaneously in the same sentence in all text files in the local area network, the correction weights of the first word and the second word are assigned a value of 0.

[0063] When the number of document co-occurrences of the first word and the second word is relatively smaller than the total number of document co-occurrences, and the document co-occurrence length of the first word and the second word is smaller than the average document length, the degree of adjustment of the weight of the edge between the first word and the second word is smaller, and the difference between the modified weight and the weight of the first word and the second word is smaller, thereby avoiding the influence of accidental local co-occurrence on the weight calculation.

[0064] Furthermore, by combining the positional relevance and semantic relevance of different words in the word segmentation results of the search text, the comprehensive relevance of different words in the word segmentation results of the search text is evaluated to avoid the deviation of semantic relevance evaluation caused by single reliance on semantic generalization. At the same time, it breaks through the traditional search's reliance on literal matching.

[0065] The semantic weights of the first word and the second word are determined according to the position relevance and the modified weights of the first word and the second word.

[0066] Preferably, as an embodiment of the present application, the average of the positional relevance and the modified weight of the first word and the second word is recorded as the semantic weight of the first word and the second word.

[0067] The same method can be used to obtain the semantic weights of any two different words in the word segmentation results of the search text.

[0068] The sentence vector of the search text is determined based on the modified weights between all words in the word segmentation results of the search text and the word vectors of all words.

[0069] The normalized value of the cumulative sum of the modified weights between the first word and all other words in the word segmentation results of the search text is recorded as the total semantic weight of the first word, and the product of the total semantic weight of the first word and the word vector is recorded as the weighted cumulative value of the first word; the cumulative sum of the weighted cumulative values ​​of all words in the word segmentation results of the search text is recorded as the sentence vector of the search text.

[0070] It should be noted that this embodiment uses the Z-Score standard normalization method to calculate the normalized value. In actual application, the implementer can use other existing methods such as maximum and minimum value normalization method, sigmoid function, etc. to calculate the normalized value, which is not limited here.

[0071] According to the method of determining the sentence vector of the search text based on the search text, the sentence vector of each sentence in all the text files in the local area network is determined based on all the text files in the local area network.

[0072] At this point, the sentence vector of each sentence in the retrieved text and all text files in the local area network is obtained.

[0073] Step S003: Determine the file vector of each text file in the local area network based on the number of times all different words in each sentence in all text files in the local area network appear simultaneously in the same sentence in all text files in the local area network, and the sentence vector of each sentence in all text files in the local area network.

[0074] Furthermore, the importance of each sentence in all text files in the local area network relative to the text files is determined.

[0075] According to the number of times all different words in each sentence in all text files in the local area network appear simultaneously in the same sentence in all text files in the local area network, and the sentence vector of each sentence in all text files in the local area network, the file vector of each text file in the local area network is determined respectively.

[0076] Any sentence in all text files in the local area network is recorded as the target file sentence, the cumulative sum of the document co-occurrence times of all different words in the word segmentation results of the target file sentence is recorded as the co-occurrence frequency of the target file sentence, the ratio of the co-occurrence frequency of the target file sentence to the total number of document co-occurrences is recorded as the fourth ratio of the target file sentence, and the product of the sentence vector of the target file sentence and the fourth ratio is recorded as the weighted vector of the target file sentence; the vector sum of the weighted vectors of all sentences in the same text file in the local area network is recorded as the file vector of the same text file in the local area network.

[0077] At this point, the file vector of each text file in the local area network is obtained.

[0078] Step S004: Complete the semantic association retrieval of digital archives based on the similarity between the archive vectors of all text archives in the local area network and the sentence vectors of the search text.

[0079] The absolute value of the cosine similarity between the sentence vector of the search text and the file vector of the text file in the local area network is recorded as the semantic association degree between the search text and the text file.

[0080] The word vectors in this embodiment are obtained using the Word2vec model. Therefore, all word vectors obtained are of the same preset length, which ensures that the calculation of cosine similarity is meaningful.

[0081] It's understandable that every text file in the local area network has a corresponding file vector. In other words, the search text has a corresponding degree of semantic relevance to every text file in the local area network. The greater the degree of semantic relevance between the search text and the text file, the more likely the text file is the corresponding search result.

[0082] The largest semantic correlation among the retrieved texts The text files corresponding to the semantic association degree are taken as the search results, and the text files are arranged in descending order according to the semantic association degree corresponding to the text files.

[0083] in, represents a preset parameter. In this embodiment, the value of the preset parameter is 10.

[0084] At this point, the semantic association retrieval of digital archives is completed.

[0085] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A semantic association retrieval method for local area network digital archives, characterized in that: The method comprises the following steps: Perform semantic association search on all text files in the local area network, collect the user's search text, and obtain the word vector of each word in the word segmentation results of the search text; Record any two different words in the word segmentation results of the search text as the first word and the second word respectively; determine the positional relevance of the first word and the second word based on the positional difference between the first word and the second word in the search text; determine the modified weights of the first word and the second word based on the sentence length and number of times the first word and the second word appear simultaneously in the same sentence in all text files in the local area network, as well as the semantic relevance of the first word and the second word; determine the semantic weights of the first word and the second word based on the positional relevance and the modified weight; determine the sentence vector of the search text based on the modified weights between all words in the word segmentation results of the search text and the word vectors of all words; and determine the sentence vector of each sentence in all text files in the local area network; Determine a file vector for each text file in the local area network according to the number of times all different words in each sentence in all text files in the local area network co-appear in the same sentence in all text files in the local area network and the sentence vector of each sentence in all text files in the local area network; According to the similarity between the file vectors of all text files in the local area network and the sentence vectors of the search text, the semantic association retrieval of digital files is completed; The method of determining the modified weights of the first word and the second word based on the length and number of times the first word and the second word appear simultaneously in the same sentence in all text files in the local area network, and the semantic relevance between the first word and the second word, includes the following specific methods: Use the TextRank algorithm to obtain the weight of the edge between the first word and the second word; When a first word and a second word appear simultaneously in the same sentence in all text files in the local area network, the first word and the second word are recorded as co-occurring in the text files; a first ratio of the first word to the second word is determined based on the difference between the number of co-occurrences of the first word and the second word and the number of co-occurrences of all different words in all text files in the local area network; determining a second ratio of the first word to the second word based on the lengths of all sentences in all text files in the local area network and the lengths of sentences in which the first word and the second word co-occur in the text files; The average of the first ratio and the second ratio of the first word to the second word is recorded as the third ratio of the first word to the second word; The difference between the weight of the edge between the first word and the second word and the third ratio is recorded as the modified weight of the first word and the second word.

2. The semantic association retrieval method for local area network digital archives according to claim 1, characterized in that: The method for determining the positional relevance between the first word and the second word is: The absolute value of the difference between the sequence numbers of the first word and the second word is recorded as the position distance between the first word and the second word, and the negative correlation processing result of the position distance between the first word and the second word is recorded as the position relevance between the first word and the second word.

3. The semantic association retrieval method for local area network digital archives according to claim 1, characterized in that: The method of determining the first ratio of the first word to the second word based on the difference between the number of co-occurrences of the first word and the second word and the number of co-occurrences of all different words in all text files in the local area network includes the following specific methods: The total number of times the first word and the second word co-occur in all text files in the local area network is recorded as the document co-occurrence number of the first word and the second word, the total number of times all different words co-occur in all text files in the local area network is recorded as the total document co-occurrence number, and the ratio of the document co-occurrence number of the first word and the second word to the total document co-occurrence number is recorded as the first ratio of the first word and the second word.

4. The semantic association retrieval method for local area network digital archives according to claim 3, characterized in that: The lengths of all sentences in all text files in the local area network, and the lengths of sentences in which the first word and the second word co-occur in the text files, are used to determine a second ratio of the first word to the second word, including the following specific method: The average length of the sentences in which the first word and the second word co-occur in the text file is recorded as the document co-occurrence length of the first word and the second word; the average length of all sentences in all text files in the local area network is recorded as the average document length; the ratio of the document co-occurrence length of the first word and the second word to the average document length is recorded as the second ratio of the first word and the second word.

5. The semantic association retrieval method for local area network digital archives according to claim 1, characterized in that: The method for determining the semantic weights of the first word and the second word is: The average of the positional relevance and the modified weight of the first word and the second word is recorded as the semantic weight of the first word and the second word.

6. The semantic association retrieval method for local area network digital archives according to claim 1, characterized in that: The method for determining the sentence vector of the search text is: The normalized value of the cumulative sum of the modified weights between the first word and all other words in the word segmentation results of the search text is recorded as the total semantic weight of the first word, and the product of the total semantic weight of the first word and the word vector is recorded as the weighted cumulative value of the first word; The sum of the weighted cumulative values ​​of all words in the word segmentation results of the search text is recorded as the sentence vector of the search text.

7. The semantic association retrieval method for local area network digital archives according to claim 3, characterized in that: The method for determining the file vector of the text file is: Record any sentence in all text files in the local area network as a target file sentence, record the cumulative sum of the document co-occurrence times of all different words in the word segmentation results of the target file sentence as the co-occurrence frequency of the target file sentence, record the ratio of the co-occurrence frequency of the target file sentence to the total number of document co-occurrences as the fourth ratio of the target file sentence, and record the product of the sentence vector of the target file sentence and the fourth ratio as the weighted vector of the target file sentence; The vector sum of the weighted vectors of all sentences in the same text file in the local area network is recorded as the file vector of the same text file in the local area network.

8. The semantic association retrieval method for local area network digital archives according to claim 1, characterized in that: The method of completing the semantic association retrieval of digital archives based on the similarity between the archive vectors of all text archives in the local area network and the sentence vectors of the search text includes the following specific methods: The absolute value of the cosine similarity between the sentence vector of the search text and the file vector of the text file in the local area network is recorded as the semantic correlation degree between the search text and the text file; According to the semantic relevance between the search text and all text files in the local area network, the semantic relevance search results of the digital files are obtained.

9. The semantic association retrieval method for local area network digital archives according to claim 8, characterized in that: The specific method of obtaining the semantic association search results of digital archives based on the semantic association degree between the search text and all text archives in the local area network includes: The largest degree of semantic relevance among all the semantic relevances corresponding to the retrieved text The text files corresponding to the semantic association degrees are arranged in descending order according to the semantic association degrees corresponding to the text files to obtain the semantic association retrieval results, among which, Indicates preset parameters.

Images