Literature information retrieval and analysis system and method based on artificial intelligence

Through artificial intelligence analysis of the subject word distribution vector and comparison vector in literature, the problem of difficulty in accurately comparing the correlation degree of subject word between different documents in the prior art is solved, and a more accurate document information retrieval effect is achieved.

CN118916459APending Publication Date: 2024-11-08HANGZHOU QIANJIANG DATA TECHNOLOGY CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410959610.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-17
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

In the prior art, it is difficult to accurately compare the correlation between subject words between different documents in literature information retrieval, resulting in a large number of documents with low correlation in the search results.

Method used

Using an artificial intelligence-based method, by obtaining the distribution vectors and comparison vectors of subject words in the literature, analyzing the distribution of subject words in different paragraphs, and calculating the differences in subject words between different documents to describe the similarities between documents more accurately. degree and degree of difference.

Benefits of technology

It improves the accuracy of document information retrieval, can describe the similarity and differences between different documents more accurately, and reduces the low correlation literature in the search results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118916459A_ABST
    Figure CN118916459A_ABST
Patent Text Reader

Abstract

The invention discloses a literature information retrieval analysis system and method based on artificial intelligence, and relates to the technical field of computer information retrieval. Subject terms in literatures are obtained, a word combination formed by the subject terms is marked as a first target combination, all the first target combinations in the literatures are marked, and the first target combinations are marked; obtaining a second target combination formed by the subject terms, counting the distribution of the subject terms in each paragraph of the literature to obtain a distribution vector of the subject terms in each paragraph, representing the distribution vector of the target combination in the same literature through a reference vector group to obtain a comparison vector of the target combination, and storing the comparison vector of the target combination in the literature; and obtaining a plurality of literatures including the target combination, comparing the literatures with the reference literatures, and comparing the comparison vectors of the reference literatures with the comparison vectors of the plurality of literatures to obtain differences of the subject terms in different literatures.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of computer information retrieval, and in particular to a document information retrieval and analysis system and method based on artificial intelligence. Background Art

[0002] With the continuous advancement of computer technology and network technology, the methods of document information retrieval and analysis are becoming increasingly diverse. In the prior art, the common method of document information retrieval is to obtain keywords from the document, connect the keywords through Boolean logic operators to form words or phrases, and further retrieve and analyze the document content through the words or phrases.

[0003] In the prior art, the relevance of several documents is obtained by comparing whether the words or phrases composed of keywords exist in the document information. However, the same words or phrases may exist in different documents in the same field, but due to the different themes of the documents, the amount of information involved in the content described by the words or phrases in different documents is different. Therefore, the traditional method can only retrieve documents including the same words or phrases, but cannot compare the relationship between the same words or phrases and the subject of the document. Therefore, in the retrieval process, a large number of documents with low relevance to the documents will be obtained. Summary of the invention

[0004] The purpose of the present invention is to provide a document information retrieval and analysis system and method based on artificial intelligence to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solution: a literature information retrieval and analysis method based on artificial intelligence:

[0006] Step S100: Obtain subject words in the document, record a word combination consisting of the subject words as a first target combination, and mark all first target combinations in the document;

[0007] Step S200: record the word combination including the subject word in the word combination and different from the first combination as the second target combination, and perform statistics on the distribution of the subject word in each paragraph of the document to obtain the distribution vector of the subject word in each paragraph;

[0008] Step S300: obtaining a reference vector group, representing the distribution vector of the target combination in the same document through the reference vector group, and obtaining a comparison vector of the target combination;

[0009] Step S400: obtaining a document from a plurality of documents as a reference document, wherein each document in the plurality of documents includes a target combination, and comparing the comparison vector of the reference document with the comparison vectors of the plurality of documents to obtain the difference in the subject words in different documents.

[0010] Furthermore, step S100 includes:

[0011] Step S101: Obtain references, obtain subject words of the references, and obtain all word combinations consisting of subject words in the references;

[0012] Step S102: obtaining a word combination from the word combinations as a first target combination, and marking the first target combination of each paragraph in the reference.

[0013] Further, step S200 includes:

[0014] Step S201: Obtain a word combination consisting of subject words from the document, and select a second target combination from the word combination, wherein the second target combination includes the subject words, and the second target combination is different from the first target combination;

[0015] Step S202: In a certain paragraph of the reference, statistics are performed on the first target combinations, wherein the statistical parameters include: the number of the first target combinations, the total number of the first target combinations and the second target combinations, and the ratio of the number of the first target combinations to the number of the second target combinations;

[0016] The first target combination represents a word or phrase composed of a certain subject word, and the second target combination represents another word or phrase composed of a certain subject word. The first target combination and the second target combination both include a certain subject word. The first target combination and the second target combination are statistically analyzed to distinguish the distribution of the first target combination.

[0017] This distribution shows the proportion of different usages of the subject word in the document. When the ratio of the first target combination to the second target combination is large, it means that the main usage of the subject word in the document is the first target combination; the meaning contained in the first target combination is closer to the content contained in the main theme of the document than the meaning contained in the second target combination;

[0018] Step S203: Obtain corresponding values ​​of at least two statistical parameters to form a distribution vector of the subject word in the certain paragraph, wherein a distribution vector α, α(q1, q2, q3, ..., q e ), where q1, q2, q3, ... and q e They represent the corresponding values ​​of the first, second, third, … and e-th statistical parameters respectively;

[0019] Step S204: obtain the number of paragraphs of the reference document, record it as n1, collect the n1 distribution vectors of the subject words, and form the first feature vector group of the subject words.

[0020] Furthermore, step S300 includes:

[0021] Step S301: obtaining an m2-dimensional reference vector group, wherein the reference vector group contains at least m1 linearly independent vectors, satisfying the condition m2≥m1≥n1;

[0022] Through m1 linearly independent vectors, at most an m1-dimensional linear space can be formed, and n1 vectors can be represented in this m1-dimensional linear space, that is, the first eigenvector group is mapped to the m1-dimensional linear space;

[0023] Step S302: Obtain the i-th distribution vector α from the first feature vector group of the reference i , by referring to the vector in the vector group, the distribution vector α i , α i =x i1 β1+x i2 β2+x i3 β3+……+x im1 β m1 , where β1, β2, β3,...and β m1 , respectively represent the first, second, third, ... and m1th vectors in the reference vector group, x i1 、x i2 、x i3 , ... and x im1 , representing β1, β2, β3, ... and β m1 The coefficients of the collection coefficients form the coefficient vector μ corresponding to the i-th distribution vector i , μ i (x i1 , x i2 , x i3 , ..., x im1 );

[0024] Step S303: The reference vector group represents all the distribution vectors in the first feature vector group of the reference document, obtains the corresponding coefficient vector of the distribution vector, and obtains the comparison vector γ of the reference document, γ=μ1+μ2+μ3+……+μ n1 , where μ1, μ2, μ3,...and μ n1 They respectively represent the coefficient vectors corresponding to the first paragraph, the second paragraph, the third paragraph, ... and the n1th paragraph in the reference.

[0025] Furthermore, step S400 includes:

[0026] Step S401: obtaining two documents from a number of documents, which are respectively recorded as a first target document and a second target document, obtaining a comparison vector of the first target document as γ1, and obtaining a comparison vector of the second target document as γ2;

[0027] Step S402: Calculate the difference between γ and γ1 and record it as diff1; calculate the difference between γ and γ2 and record it as diff2;

[0028] In the literature of the same field, the subject terms involved are similar, but different documents have different descriptions or expressions of the same subject term, or different subject terms contribute to the main theme of the document to different degrees, resulting in different degrees of expression of the same subject term in different documents. Such differences are reflected in the differences in the vocabulary or phrases composed of the same subject term in different documents;

[0029] Step S403: when diff1<diff2, it is determined that when the subject word is used as the reference object, the similarity between the first target document and the reference document is greater than the similarity between the second target document and the reference document;

[0030] Furthermore, when several documents respectively include K kinds of subject words, the comparison method between the reference document and the first target document includes:

[0031] Step S51: Obtain the number of times various subject terms appear in the reference document and the first target document respectively, wherein the first subject term appears P01 times, the second subject term appears P02 times, ... and the Kth subject term appears P0K times in the reference document, and the first subject term appears P11 times, the second subject term appears P12 times, ... and the Kth subject term appears P1K times in the first reference document;

[0032] Step S52: When various subject words of the reference document and the first target document are obtained as reference objects, the difference of the comparison vectors is compared. The difference of the comparison vector of the first subject word, the difference of the comparison vector of the second subject word, the difference of the comparison vector of the third subject word, ... and the difference of the comparison vector of the Kth subject word are respectively recorded as diff 1 , diff 2 , diff 3 , ... and diff K ;

[0033] Step S53: Calculate the difference D of K kinds of subject words K , Among them, P0r represents the number of times the r-th subject word appears in the reference, P1r represents the number of times the r-th subject word appears in the first target document, and diff rrepresents the difference between the comparison vectors of the r-th topic word, where r≤K;

[0034] The frequencies of the same subject terms in the two documents are compared, and the proportional coefficient of the difference is calculated by the proportion of the changes in the same subject terms in the two documents to the references. When the frequency of the subject terms in the two documents differs greatly, the difference of the comparison vector is further improved to make the difference of the documents more prominent.

[0035] In order to better implement the above method, a document information retrieval and analysis system based on artificial intelligence document information retrieval and analysis method is also proposed;

[0036] The system includes:

[0037] A subject word management module, a distribution vector management module, a comparison vector management module and a comparison module, wherein the subject word management module is used to manage word combinations including subject words in the document, the distribution vector management module is used to manage the distribution vectors of subject words in each paragraph, the comparison vector management module is used to manage the comparison vectors, and the comparison module is used to compare the differences of the same subject word in different documents;

[0038] Furthermore, the subject word management module includes: a subject word acquisition unit and a word combination marking unit, wherein the subject word acquisition unit is used to acquire subject words in the document, and the word combination marking unit is used to mark the word group including the subject words in the document;

[0039] Further, the distribution vector management module includes: a word combination management unit, a word combination statistics unit, a distribution vector management unit and a feature vector group management unit, wherein the word combination management unit is used to manage the first target combination and the second target combination, the word combination statistics unit is used to count the word combinations, the distribution vector management unit is used to obtain the distribution vector corresponding to the statistical parameter composition paragraph, and the feature vector group management unit is used to manage the first feature vector group;

[0040] Further, the comparison vector management module includes: a reference vector group management unit, a vector representation unit and a comparison vector generation unit, wherein the reference vector group management unit is used to manage the reference vector group, the vector representation unit is used to represent the vectors in the first feature vector group by the vectors in the reference vector group, and the comparison vector generation unit is used to generate the comparison vector corresponding to the subject word;

[0041] Furthermore, the comparison module includes: a comparison vector acquisition unit, a difference calculation unit, a difference degree comparison unit and a coefficient calculation unit, wherein the comparison vector acquisition unit is used to obtain comparison vectors of different documents, the difference calculation unit is used to calculate the difference degree of the comparison vectors, the difference degree comparison unit is used to judge the calculation result of the difference degree of the comparison vectors, and the coefficient calculation unit is used to calculate the corresponding coefficient of the difference degree of the comparison vectors.

[0042] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: the present invention analyzes the subject words in each paragraph of the document, constructs the distribution vector of the subject words in different paragraphs according to the distribution of the subject words in each paragraph, and further analyzes the role of the subject words in the document by extracting the distribution of the subject words in different paragraphs. On the basis of the prior art analysis of whether the words or phrases composed of the subject words exist, the distribution of the words or phrases in the document is further analyzed, and the characteristics of the distribution are extracted through the distribution vector. When comparing the similarity of the documents, the characteristics of the distribution of the subject words are added, so that the similarity and difference between different documents can be described more accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0044] Figure 1 It is a structural schematic diagram of a document information retrieval and analysis system based on artificial intelligence of the present invention;

[0045] Figure 2 It is a flow chart of a method for document information retrieval and analysis based on artificial intelligence of the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] See also Figure 1 and Figure 2 , the present invention provides a technical solution:

[0048] A literature information retrieval and analysis method based on artificial intelligence:

[0049] Step S100: Obtain subject words in the document, record a word combination consisting of the subject words as a first target combination, and mark all first target combinations in the document;

[0050] Wherein, step S100 includes:

[0051] Step S101: Obtain references, obtain subject words of the references, and obtain all word combinations consisting of subject words in the references;

[0052] Step S102: obtaining a word combination from the word combinations as a first target combination, and marking the first target combination of each paragraph in the reference;

[0053] In an embodiment, the subject words of the document are preferably obtained by an artificial intelligence algorithm using an automated method, for example:

[0054] Keyword extraction methods based on deep learning, such as the BERT model, BERT stands for Bidirectional Encoder Representations from Transformers. BERT aims to pre-train deep bidirectional representations by jointly conditioning on the left and right contexts in all layers;

[0055] Keyword extraction methods based on word vectors, such as the Word2Vec model. Word2Vec is a related model used to generate word vectors. The Word2Vec model can be used to map each word to a vector, which can be used to represent the relationship between words.

[0056] Step S200: record the word combination including the subject word in the word combination and different from the first combination as the second target combination, and perform statistics on the distribution of the subject word in each paragraph of the document to obtain the distribution vector of the subject word in each paragraph;

[0057] Wherein, step S200 includes:

[0058] Step S201: Obtain a word combination consisting of subject words from the document, and select a second target combination from the word combination, wherein the second target combination includes the subject words, and the second target combination is different from the first target combination;

[0059] Step S202: In a certain paragraph of the reference, statistics are performed on the first target combinations, wherein the statistical parameters include: the number of the first target combinations, the total number of the first target combinations and the second target combinations, and the ratio of the number of the first target combinations to the number of the second target combinations;

[0060] For example, take subject word a. In paragraph par1, the words that include subject word a are C1(a, b) and C2(a, c), where b and c represent the words that make up words C1 and C2 respectively.

[0061] Take vocabulary C1 as the first target combination and vocabulary C2 as the second target combination;

[0062] Among them, in paragraph par1, there are 4 in vocabulary C1 and 3 in vocabulary C2;

[0063] The proportion of vocabulary C1 is 4÷(4+3)=0.571, and the ratio of vocabulary C1 to vocabulary C2 is 1.3;

[0064] Construct the distribution vector α1 of the topic word a in paragraph par1: (4, 0.57, 1.3)

[0065] Step S203: Obtain corresponding values ​​of at least two statistical parameters to form a distribution vector of the subject word in the certain paragraph, wherein a distribution vector α, α(q1, q2, q3, ..., q e ), where q1, q2, q3, ... and q e They represent the corresponding values ​​of the first, second, third, … and e-th statistical parameters respectively;

[0066] Step S204: obtaining the number of paragraphs of the reference document as n1, and collecting the n1 distribution vectors of the subject words to form a first feature vector group of the subject words;

[0067] Step S300: obtaining a reference vector group, representing the distribution vector of the target combination in the same document through the reference vector group, and obtaining a comparison vector of the target combination;

[0068] Wherein, step S300 includes:

[0069] Step S301: obtaining an m2-dimensional reference vector group, wherein the reference vector group contains at least m1 linearly independent vectors, satisfying the condition m2≥m1≥n1;

[0070] Step S302: Obtain the i-th distribution vector α from the first feature vector group of the reference i , by referring to the vector in the vector group, the distribution vector α i , α i =x i1 β1+x i2 β2+x i3 ·β3+……+xi m1 β m1 , where β1, β2, β3,...and β m1, respectively represent the first, second, third, ... and m1th vectors in the reference vector group, x i1 、x i2 、x i3 , ... and x im1 , representing β1, β2, β3, ... and β m1 The coefficients of the collection coefficients form the coefficient vector μ corresponding to the i-th distribution vector i , μ i (x i1 , x i2 , x i3 , ..., x im1 );

[0071] Get the distribution vector α2 (2, 0.286, 0.4) of the topic word a in paragraph par2, and get the distribution vector α3 (3, 0.75, 3) of the topic word a in paragraph par3;

[0072] The distribution vector of the paragraph is represented by a 3-dimensional vector group β1(1,0,0), β2(0,1,0) and β3(0,0,1);

[0073] α1=4·β1+0.57·β2+1.3·β3, and we get the coefficient vector (4, 0.57, 1.3);

[0074] By adjusting the values ​​of the reference vector group, the total weight of different statistical parameters is adjusted, for example, β1 is adjusted to β'1, β2 and β3 remain unchanged, where β'1 (2, 0, 0);

[0075] The adjusted distribution vector α'1(2, 0.57, 1.3) reduces the weight of the first statistical parameter;

[0076] Step S303: The reference vector group represents all the distribution vectors in the first feature vector group of the reference document, obtains the corresponding coefficient vector of the distribution vector, and obtains the comparison vector γ of the reference document, γ=μ1+μ2+μ3+……+μ n1 , where μ1, μ2, μ3,...and μ n1 Respectively represent the coefficient vectors corresponding to the first paragraph, the second paragraph, the third paragraph, ... and the n1th paragraph in the reference;

[0077] Step S400: obtaining a document from a plurality of documents as a reference document, each of the plurality of documents including the target combination, and comparing the comparison vector of the reference document with the comparison vectors of the plurality of documents to obtain the difference of the subject words in different documents;

[0078] Wherein, step S400 includes:

[0079] Step S401: obtaining two documents from a number of documents, which are respectively recorded as a first target document and a second target document, obtaining a comparison vector of the first target document as γ1, and obtaining a comparison vector of the second target document as γ2;

[0080] Step S402: Calculate the difference between γ and γ1 and record it as diff1; calculate the difference between γ and γ2 and record it as diff2;

[0081] The difference calculation methods include: Euclidean distance, Manhattan distance or cosine similarity;

[0082] Among them, in the calculation method of Euclidean distance, the vector is regarded as a point in a high-dimensional space, and the difference of the vector is obtained by calculating the distance between two points;

[0083] In the calculation method of Manhattan distance, in the spatial coordinate system, the sum of the absolute wheel distances of two points in the standard coordinate system;

[0084] In the calculation method of cosine similarity, the similarity between two vectors is measured by measuring the cosine value of the angle between them. Cosine similarity reflects the difference in direction of the vectors, not the length or distance;

[0085] Step S403: when diff1<diff2, it is determined that when the subject word is used as the reference object, the similarity between the first target document and the reference document is greater than the similarity between the second target document and the reference document;

[0086] Among them, when several documents include K kinds of subject words respectively, the comparison method between the reference document and the first target document includes:

[0087] Step S51: Obtain the number of times various subject terms appear in the reference document and the first target document respectively, wherein the first subject term appears P01 times, the second subject term appears P02 times, ... and the Kth subject term appears P0K times in the reference document, and the first subject term appears P11 times, the second subject term appears P12 times, ... and the Kth subject term appears P1K times in the first reference document;

[0088] Step S52: When various subject words of the reference document and the first target document are obtained as reference objects, the difference of the comparison vectors is compared. The difference of the comparison vector of the first subject word, the difference of the comparison vector of the second subject word, the difference of the comparison vector of the third subject word, ... and the difference of the comparison vector of the Kth subject word are respectively recorded as diff 1 , diff 2 , diff 3 , ... and diff K ;

[0089] Step S53: Calculate the difference D of K kinds of subject words K , Among them, P0r represents the number of times the r-th subject word appears in the reference, P1r represents the number of times the r-th subject word appears in the first target document, and diff r Represents the difference between the r-th topic word comparison vector, where r≤K.

[0090] The system includes:

[0091] Theme word management module, distribution vector management module, comparison vector management module and comparison module;

[0092] The subject word management module is used to manage word combinations including subject words in the document, wherein the subject word management module includes: a subject word acquisition unit and a word combination marking unit, wherein the subject word acquisition unit is used to acquire subject words in the document, and the word combination marking unit is used to mark word groups including subject words in the document;

[0093] The distribution vector management module is used to manage the distribution vectors of the subject words in each paragraph, wherein the distribution vector management module includes: a word combination management unit, a word combination statistics unit, a distribution vector management unit and a feature vector group management unit, wherein the word combination management unit is used to manage the first target combination and the second target combination, the word combination statistics unit is used to count the word combinations, the distribution vector management unit is used to obtain the distribution vector corresponding to the statistical parameter composition paragraph, and the feature vector group management unit is used to manage the first feature vector group;

[0094] The comparison vector management module is used to manage the comparison vectors, wherein the comparison vector management module includes: a reference vector group management unit, a vector representation unit and a comparison vector generation unit, wherein the reference vector group management unit is used to manage the reference vector group, the vector representation unit is used to represent the vectors in the first feature vector group by the vectors in the reference vector group, and the comparison vector generation unit is used to generate the comparison vector corresponding to the subject word;

[0095] Among them, the comparison module is used to compare the differences of the same subject word in different documents, wherein the comparison module includes: a comparison vector acquisition unit, a difference calculation unit, a difference degree comparison unit and a coefficient calculation unit, wherein the comparison vector acquisition unit is used to obtain the comparison vectors of different documents, the difference calculation unit is used to calculate the difference degree of the comparison vector, the difference degree comparison unit is used to judge the calculation result of the difference degree of the comparison vector, and the coefficient calculation unit is used to calculate the corresponding coefficient of the difference degree of the comparison vector.

[0096] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0097] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A literature information retrieval and analysis method based on artificial intelligence, characterized in that: The method comprises the following steps: Step S100: Obtain subject words in the document, record a word combination consisting of the subject words as a first target combination, and mark all first target combinations in the document; Step S200: record the word combination including the subject word in the word combination and different from the first combination as the second target combination, and perform statistics on the distribution of the subject word in each paragraph of the document to obtain the distribution vector of the subject word in each paragraph; Step S300: obtaining a reference vector group, representing the distribution vector of the target combination in the same document through the reference vector group, and obtaining a comparison vector of the target combination; Step S400: obtaining a document as a reference document from a plurality of documents, wherein each document in the plurality of documents includes a target combination, and comparing a comparison vector of the reference document with the comparison vectors of the plurality of documents to obtain differences in the subject words in different documents.

2. The method for document information retrieval and analysis based on artificial intelligence according to claim 1, characterized in that: Step S100 includes: Step S101: Obtain the reference document, obtain the subject words of the reference document, and obtain all word combinations consisting of the subject words in the reference document; Step S102: obtaining a word combination from the word combinations as a first target combination, and marking the first target combination of each paragraph in the reference.

3. The method for document information retrieval and analysis based on artificial intelligence according to claim 2, characterized in that: Step S200 includes: Step S201: Obtain a word combination consisting of subject words from the document, and select a second target combination from the word combination, wherein the second target combination includes the subject words, and the second target combination is different from the first target combination; Step S202: In a certain paragraph of the reference, statistics are performed on the first target combinations, wherein the statistical parameters include: the number of the first target combinations, the total number of the first target combinations and the second target combinations, and the ratio of the number of the first target combinations to the number of the second target combinations; Step S203: Obtain corresponding values ​​of at least two statistical parameters to form a distribution vector of the subject word in the certain paragraph, wherein a distribution vector α, α(q1, q2, q3, ..., q e ), where q1, q2, q3, ... and q e They represent the corresponding values ​​of the first, second, third, … and e-th statistical parameters respectively; Step S204: obtain the number of paragraphs of the reference document, record it as n1, collect the n1 distribution vectors of the subject words, and form a first feature vector group of the subject words.

4. The method for document information retrieval and analysis based on artificial intelligence according to claim 3, characterized in that: Step S300 includes: Step S301: obtaining an m2-dimensional reference vector group, wherein the reference vector group contains at least m1 linearly independent vectors and satisfies the condition m2≥m1≥n1; Step S302: Obtain the i-th distribution vector α from the first feature vector group of the reference i , by referring to the vector in the vector group, the distribution vector α i , α i =x i1 β1+x i2 β2+x i3 ·β3+……+xi m1 β m1 , among which, β1, β2, β3,...and β m1 , respectively represent the first, second, third, ... and m1th vectors in the reference vector group, x i1 、x i2 、x i3 , ... and x im1 , representing β1, β2, β3, ... and β m1 The coefficients of the collection coefficients form the coefficient vector μ corresponding to the i-th distribution vector i , μ i (x i1 , x i2 , x i3 , ..., x im1 ); Step S303: The reference vector group represents all the distribution vectors in the first feature vector group of the reference document, obtains the corresponding coefficient vector of the distribution vector, and obtains the comparison vector γ of the reference document, γ=μ1+μ2+μ3+……+μ n1 , where μ1, μ2, μ3,...and μ n1 They respectively represent the coefficient vectors corresponding to the first paragraph, the second paragraph, the third paragraph, ... and the n1th paragraph in the reference.

5. The method for document information retrieval and analysis based on artificial intelligence according to claim 4, characterized in that: Step S400 includes: Step S401: obtaining two documents from the plurality of documents, which are respectively recorded as a first target document and a second target document, obtaining a comparison vector of the first target document as γ1, and obtaining a comparison vector of the second target document as γ2; Step S402: Calculate the difference between γ and γ1 and record it as diff1; calculate the difference between γ and γ2 and record it as diff2; Step S403: When diff1<diff2, it is determined that when the subject word is used as a reference object, the similarity between the first target document and the reference document is greater than the similarity between the second target document and the reference document.

6. The method for document information retrieval and analysis based on artificial intelligence according to claim 4, characterized in that: When the several documents respectively include K kinds of subject words, the comparison method between the reference document and the first target document includes: Step S51: Obtain the number of times various subject terms appear in the reference document and the first target document respectively, wherein the first subject term appears P01 times, the second subject term appears P02 times, ... and the Kth subject term appears P0K times in the reference document, and the first subject term appears P11 times, the second subject term appears P12 times, ... and the Kth subject term appears P1K times in the first reference document; Step S52: When various subject words of the reference document and the first target document are obtained as reference objects, the difference of the comparison vectors is compared. The difference of the comparison vector of the first subject word, the difference of the comparison vector of the second subject word, the difference of the comparison vector of the third subject word, ... and the difference of the comparison vector of the Kth subject word are respectively recorded as diff 1 , diff 2 , diff 3 , ... and diff K ; Step S53: Calculate the difference D of the K types of subject words K , Among them, P0r represents the number of times the r-th subject word appears in the reference, P1r represents the number of times the r-th subject word appears in the first target document, and diff r Represents the difference between the r-th topic word comparison vector, where r≤K.

7. A document information retrieval and analysis system based on artificial intelligence, used to execute a document information retrieval and analysis method based on artificial intelligence as claimed in any one of claims 1 to 6, characterized in that: The system includes the following modules: a subject word management module, a distribution vector management module, a comparison vector management module and a comparison module, wherein the subject word management module is used to manage word combinations including subject words in documents, the distribution vector management module is used to manage distribution vectors of subject words in various paragraphs, the comparison vector management module is used to manage comparison vectors, and the comparison module is used to compare the differences of the same subject word in different documents.

8. The document information retrieval and analysis system based on artificial intelligence according to claim 7, characterized in that: The subject word management module includes: a subject word acquisition unit and a word combination marking unit, wherein the subject word acquisition unit is used to acquire subject words in the document, and the word combination marking unit is used to mark the word group including the subject words in the document; The distribution vector management module includes: a word combination management unit, a word combination statistics unit, a distribution vector management unit and a feature vector group management unit, wherein the word combination management unit is used to manage the first target combination and the second target combination, the word combination statistics unit is used to count the word combinations, the distribution vector management unit is used to obtain the distribution vector corresponding to the statistical parameter composition paragraph, and the feature vector group management unit is used to manage the first feature vector group.

9. The document information retrieval and analysis system based on artificial intelligence according to claim 7, characterized in that: The comparison vector management module includes: a reference vector group management unit, a vector representation unit and a comparison vector generation unit, wherein the reference vector group management unit is used to manage the reference vector group, the vector representation unit is used to represent the vectors in the first feature vector group by the vectors in the reference vector group, and the comparison vector generation unit is used to generate the comparison vector corresponding to the subject word.

10. The document information retrieval and analysis system based on artificial intelligence according to claim 7, characterized in that: The comparison module includes: a comparison vector acquisition unit, a difference calculation unit, a difference comparison unit and a coefficient calculation unit, wherein the comparison vector acquisition unit is used to obtain comparison vectors of different documents, the difference calculation unit is used to calculate the difference of the comparison vectors, the difference comparison unit is used to judge the calculation result of the difference of the comparison vectors, and the coefficient calculation unit is used to calculate the corresponding coefficient of the difference of the comparison vectors.