A method and system for automatically analyzing article logical relations

Abstract

The embodiment of the present application discloses a method and system for automatically extracting text semantic signals, which belongs to the technical field of natural language processing. The method includes inputting an article text or a paragraph text to be analyzed, segmenting the input text, segmenting the text into multiple sentences, analyzing the grammatical structure and context information in a target sentence, identifying the logical connection relationship between the target sentence and adjacent sentences, obtaining a logical relationship score, calculating the semantic relevance between the target sentence and the adjacent sentences according to the semantic content of the target sentence and the adjacent sentences, generating a relevance score according to the semantic relevance calculation, calculating the logical coherence score of the entire text by combining the logical relationship score and the relevance score, obtaining an analysis result, and displaying the logical coherence score, the logical relationship score and the relevance score. The embodiment of the present application can help users judge the logical clarity and coherence of an article, quickly process and analyze large-scale text data, and save a lot of time and energy.

Description

Technical Field

[0001] The present application belongs to the technical field of natural language processing, and specifically, relates to a method and system for automatically analyzing the logical relationship of an article. Background Art

[0002] With the vigorous development of the Internet and the rapid growth of data volume, text analysis technology has been increasingly widely used. Text analysis technology is playing an increasingly important role, and more and more people have put forward demands for text analysis.

[0003] In the field of natural language processing, the technology of analyzing the logical relationship of articles has made certain progress. For example, text entailment technology relies on a large amount of training text and manual labeling, but it consumes a lot of labor costs. The traditional method that only relies on simple keyword matching is prone to misjudgment or omission when faced with a variety of expressions. Existing methods are usually unable to dynamically adjust the analysis strategy according to the complexity of the text content, and often use fixed analysis models, which makes it difficult to achieve consistent results in different types of texts. At the same time, the judgment of the relationship between sentences has no direct connection with general text analysis. The judgment of the attitude of the sentence is simply based on the literal meaning of the word, and does not deeply consider the influence of the specific position of the word in the sentence, nor does it consider whether the current position of the word really has meaning. Summary of the invention

[0004] In order to solve the above problems and technical defects, the embodiment of the present application adopts the following technical solution, a method for automatically analyzing the logical relationship of an article, comprising the following steps: Step 1: Input the article text or paragraph text to be analyzed, and segment the input text into multiple sentences; Step 2: Analyze the grammatical structure and context information in the target sentence, identify the logical connection relationship between the target sentence and adjacent sentences, and obtain the logical relationship score; Step 3: Calculate the semantic relevance between the target sentence and the adjacent sentences according to the semantic contents of the target sentence and the adjacent sentences, and generate a relevance score according to the semantic relevance calculation; Step 4: Combine the logical relationship score and the relevance score to calculate the logical coherence score of the entire text and obtain the analysis result; Step 5: Display the logical coherence score, logical relationship score and relevance score.

[0005] Preferably, the steps of identifying the logical relationship are as follows: Obtain excellent historical essays, count the logical relationships of the excellent historical essays, and obtain the logical connection pattern of the text under each theme; Preprocess the target sentence, including word segmentation, part-of-speech tagging, and named entity recognition; Determine whether the sentence has opinions and attitudes. If so, obtain the structure and annotation information of the target sentence and extract the logical relationship features of the sentence; The target sentence is scored according to the logical connection pattern and logical relationship features to obtain a logical relationship score.

[0006] Furthermore, the determination of whether a sentence has a viewpoint and an attitude is to determine whether a sentence has a strong viewpoint and an attitude, and the determination is made based on the classification of words, the position of words, and whether the part where the words are located contains realistic meaning.

[0007] Going further, the specific process of judging whether a sentence has opinions and attitudes is as follows: Perform word segmentation on the target sentence, and traverse and judge the sentences after word segmentation; Determine whether the sentence contains strong attitude and opinion words. If so, judge that there is an attitude. If not, continue to the next round of judgment. Determine whether the sentence contains opinion nouns. If so, judge that there is an attitude. If not, proceed to the next round of judgment. Determine whether the sentence contains modal adverbs. If so, judge that it has an attitude. If not, proceed to the next round of judgment. Determine whether the sentence contains an emotional predicate verb. If so, it is judged to have an attitude. If not, continue to the next round of judgment. Determine whether the sentence contains polysex words. If so, judge that it has an attitude. If not, proceed to the next round of judgment. If there is no attitude word in the sentence, determine whether all the words in the sentence contain special phrases and structures. If so, it is judged that there is an attitude. If not, it is judged that there is no attitude.

[0008] Furthermore, before performing word segmentation on the sentences, it is necessary to establish a data vocabulary, establish multiple classification vocabulary in the data vocabulary, obtain multiple historical article texts and paragraph texts, perform word segmentation on the historical article texts and paragraph texts, store the different types of words after word segmentation in the corresponding classification vocabulary, and traverse and judge the target sentences through the words in the classification vocabulary. The classification vocabulary includes: strong attitude and opinion vocabulary, opinion noun library, modal adverb library, emotional predicate verb library, polysemic vocabulary library, special phrase library. After the traversal and judgment are completed, the words in the target sentence will be stored in the corresponding classification vocabulary.

[0009] Furthermore, the calculation steps of the relevance score are as follows: Preprocess the two sentences to be compared, perform preliminary word segmentation and part-of-speech tagging, and obtain the sentence structure, which includes: keywords, part-of-speech information, and syntactic structure; Filter keywords according to their parts of speech, retain the key parts related to semantics, and remove irrelevant parts of the sentence; Perform keyword similarity matching on the target sentence based on the sentence structure. The matching priority is based on the logical relationship score from high to low to obtain the keyword matching score; The comprehensive matching rate of the target sentence is obtained by combining the matching score with the number and proportion of keywords, and the matching mode of the target sentence is determined according to the comprehensive matching rate; The matching score of each keyword is combined with the matching pattern to obtain the relevance score of the target sentence.

[0010] Furthermore, the steps of calculating the logical coherence score of the entire text are as follows: A comprehensive logical coherence evaluation model is established. The logical relationship scores and relevance scores are normalized according to the paragraph length. The logical weights and matching weights are calculated according to the logical connection patterns and matching patterns. The logical relationship scores and relevance scores are weighted according to the logical weights and matching weights to obtain a comprehensive logical coherence score.

[0011] An automatic analysis system for article logical relations, comprising: The text segmentation module is used to input the article text or paragraph text to be analyzed, and segment the input text into multiple sentences; The logic scoring module is used to analyze the grammatical structure and context information in the target sentence, identify the logical connection relationship between the target sentence and adjacent sentences, and obtain the logical relationship score; A relevance score module is used to calculate the semantic relevance between the target sentence and the adjacent sentences according to the semantic contents of the target sentence and the adjacent sentences, and generate a relevance score according to the semantic relevance calculation; A coherence score module is used to combine the logical relationship score and the relevance score to calculate the logical coherence score of the entire text and obtain the analysis result; The display module is used to display the logical coherence score, the logical relationship score and the relevance score.

[0012] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the method for automatically analyzing the logical relationship of an article as described above are implemented.

[0013] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for automatically analyzing the logical relationship of an article as described above.

[0014] Compared with the prior art, the beneficial effects of the embodiments of the present application are: (1) This application significantly improves the adaptability to articles of different styles and structures by automatically identifying multiple logical relationships when analyzing articles, rather than being limited to specific keywords or fixed patterns. It also identifies and analyzes the logical relationships between sentences within articles, generates a logical coherence score, and helps users judge the logical clarity and coherence of articles, thus solving the problem of misjudgment or missed judgment caused by traditional systems relying on fixed rules. (2) This application comprehensively considers the overall logical structure of the article and the semantic relevance between sentences to provide a more comprehensive evaluation of logical coherence. When writing or reading, the feedback provided can help users discover and optimize the structure of the article, enabling users to have a more comprehensive understanding of the logical coherence of the article, quickly process and analyze large-scale text data, and save a lot of time and energy; (3) This application can provide a more detailed method for judging the relationship between sentences, so as to accurately judge the specific role of sentences before and after, accurately grasp the attitude part of the text, and thus judge whether a sentence contains strong opinions and attitudes, providing an accurate basis for judging the specific language purpose of the subsequent entire paragraph. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In the attached picture: Figure 1 A schematic diagram of the method steps of an embodiment of the present application; Figure 2 A schematic diagram of the system structure of an embodiment of the present application; Figure 3 This is a schematic diagram of the device structure of an embodiment of the present application. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Generally, the components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations. Example

[0017] like Figure 1 As shown, a method for automatically analyzing the logical relationship of an article includes the following steps: Input the article text or paragraph text to be analyzed, segment the input text, and divide the text into multiple sentences; Analyze the grammatical structure and contextual information in the target sentence, identify the logical connection between the target sentence and adjacent sentences, and obtain the logical relationship score; The steps to identify logical relationships are as follows: Obtain excellent historical essays, count the logical relationships of the excellent historical essays, and obtain the logical connection pattern of the text under each theme; Preprocess the target sentence, including word segmentation, part-of-speech tagging, and named entity recognition; Determine whether the sentence has opinions and attitudes. If so, obtain the structure and annotation information of the target sentence and extract the logical relationship features of the sentence; The target sentence is scored according to the logical connection pattern and logical relationship features to obtain a logical relationship score.

[0018] Judging whether a sentence has opinions and attitudes is to judge whether a sentence has strong opinions and attitudes. It is based on the classification of words, the position of words, and whether the part where the words are located has real meaning. For example, some parts of the sentence may only serve as conjunctions and have no real meaning, for example: someone says ....

[0019] The specific process of judging whether a sentence has opinions and attitudes is as follows: Perform word segmentation on the target sentence, and traverse and judge the sentences after word segmentation; Determine whether the sentence contains strong attitude and opinion words. If so, it is judged that there is an attitude. If not, continue to the next round of judgment. As long as there are strong attitude and opinion words in the sentence, there must be a strong attitude. Determine whether the sentence contains a point of view noun. If so, then it is judged that there is an attitude. If not, continue to the next round of judgment. As long as the point of view noun is the subject in the main sentence, there must be an attitude. Determine whether the sentence contains modal adverbs. If yes, then it is judged that there is an attitude. If not, continue to the next round of judgment. As long as the modal adverb appears as an adverb in the main sentence, there is basically an attitude. Determine whether the sentence contains emotional predicate verbs. If yes, then it is judged that there is an attitude. If not, continue to the next round of judgment. As long as the emotional predicate verb appears as a predicate verb in the main sentence, there is basically an attitude. Determine whether the sentence contains words of multiple genders. If so, it is determined to have an attitude. If not, continue to the next round of judgment. When judging and looking for words with strong attitudes, there will be some words without attitudes. Only when the words without attitudes are of one or more parts of speech will they have attitudes. If they are not of this part of speech, they basically have no attitude. If there is no attitude word in the sentence, determine whether all the words in the sentence contain special phrases and structures. If so, it is judged that there is an attitude. If not, it is judged that there is no attitude. If there are some additional special structures or phrases, if these situations occur, it is also judged that there is an attitude.

[0020] Before performing word segmentation on a sentence, a data word library needs to be established. Multiple classification word libraries are established in the data word library, and multiple historical article texts and paragraph texts are obtained. The historical article texts and paragraph texts are segmented, and the different types of words after word segmentation are stored in the corresponding classification word library. The target sentence is traversed and judged through the words in the classification word library. The classification word library includes: a strong attitude and opinion word library, an opinion noun library, a modal adverb library, an emotional predicate verb library, a polysemic word library, and a special phrase library. After the traversal and judgment are completed, the words in the target sentence will be stored in the corresponding classification word library.

[0021] Calculating the semantic relevance between the target sentence and the adjacent sentences according to the semantic contents of the target sentence and the adjacent sentences, and generating a relevance score according to the semantic relevance calculation; The steps for calculating the relevance score are as follows: Preprocess the two sentences to be compared, perform preliminary word segmentation and part-of-speech tagging, and obtain the sentence structure, which includes: keywords, part-of-speech information, and syntactic structure; Filter keywords according to their parts of speech, retain the key parts related to semantics, and remove irrelevant parts of the sentence, such as function words; Perform keyword similarity matching on the target sentence based on the sentence structure. The matching priority is based on the logical relationship score from high to low to obtain the keyword matching score; The comprehensive matching rate of the target sentence is obtained by combining the matching score with the number and proportion of keywords, and the matching mode of the target sentence is determined according to the comprehensive matching rate; The matching score of each keyword is combined with the matching pattern to obtain the relevance score of the target sentence.

[0022] The final sentence relevance relationship is obtained by comparing the valuable words within the sentence. The calculation steps of the sentence relevance relationship are as follows: Traverse all the meaningful words in the target sentence and remove meaningless words; Traverse all the meaningful words in the adjacent sentences and remove the meaningless words; Compare all the words in the target sentence with those in the adjacent sentences to see if there is a first-level matching word. If there is a match, mark the first-level match. If not, proceed to the next step. According to the part of speech of all words in the target sentence, read the secondary matching word list to determine whether the word has a related secondary word in the adjacent sentence. If it matches, mark the secondary match. If not, proceed to the next step. According to the part of speech of all words in the target sentence, read the third-level matching word list to determine whether the word has a related third-level word in the adjacent sentence. If it matches, mark the third-level match; if not, mark no match; Finally, based on the ratio of the first-level, second-level, and third-level matches of all the words in the target sentence in the adjacent sentences, a complex rule is used to judge the relationship between the target sentence and the adjacent sentences, and a similarity label and the corresponding similarity score are output.

[0023] The mechanism for eliminating irrelevant words is based on the previous logic. The description methods of some function words, such as someone says, it ie believed that... and similar parts are considered meaningless. If the part where the word is located has this label, the word will be removed directly. In addition, only nouns and verbs are considered in the comparison.

[0024] The classification vocabulary stores different types of words. The first-level vocabulary is the original vocabulary, that is, it is compared to see if there are original words; the second-level vocabulary is the synonym vocabulary, which is a library of standard words and their related synonyms, divided into noun library and verb library; the third-level vocabulary is the domain vocabulary, all collected words are classified into domains, and divided into noun library and verb library.

[0025] The rule for judging the relationship between sentences is to first read the third-level vocabulary to determine how many words in the target sentence have original word matches, synonym matches, and domain matches in adjacent sentences, and at the same time calculate the matching rates in verbs and nouns, and pronouns are directly counted as original word matches.

[0026] Whether it is a full match or a partial match is determined based on whether there are words that are not matched. In full matches and partial matches, different labels and scores will be given based on the proportion of original word matches, synonym matches, and field matches. Basically, the higher the level of matches, the higher the score. In partial matches, a ratio is calculated based on the verb and noun matching rate multiplied by the weight. In partial matches, if the ratio is too low, it will be directly judged as a mismatch.

[0027] Combine the logical relationship score and the relevance score to calculate the logical coherence score of the entire text and obtain the analysis result; The steps to calculate the logical coherence score of the entire text are as follows: A comprehensive logical coherence evaluation model is established. The logical relationship scores and relevance scores are normalized according to the paragraph length. The logical weights and matching weights are calculated according to the logical connection patterns and matching patterns. The logical relationship scores and relevance scores are weighted according to the logical weights and matching weights to obtain a comprehensive logical coherence score.

[0028] The logical coherence score, logical relationship score and relevance score are displayed. Example

[0029] like Figure 2 As shown, an automatic analysis system for article logical relations comprises: The text segmentation module is used to input the article text or paragraph text to be analyzed, and segment the input text into multiple sentences; The logic scoring module is used to analyze the grammatical structure and context information in the target sentence, identify the logical connection relationship between the target sentence and adjacent sentences, and obtain the logical relationship score; A relevance score module is used to calculate the semantic relevance between the target sentence and the adjacent sentences according to the semantic contents of the target sentence and the adjacent sentences, and generate a relevance score according to the semantic relevance calculation; A coherence score module is used to combine the logical relationship score and the relevance score to calculate the logical coherence score of the entire text and obtain the analysis result; The display module is used to display the logical coherence score, the logical relationship score and the relevance score. Example

[0030] like Figure 3 As shown, from the hardware level, the present application provides an embodiment of an electronic device that implements all or part of the content of the method for automatically analyzing the logical relationship of an article, and the electronic device specifically includes the following content: Step 1: Input the article text or paragraph text to be analyzed, and segment the input text into multiple sentences; Step 2: Analyze the grammatical structure and context information in the target sentence, identify the logical connection relationship between the target sentence and adjacent sentences, and obtain the logical relationship score; Step 3: Calculate the semantic relevance between the target sentence and the adjacent sentences according to the semantic contents of the target sentence and the adjacent sentences, and generate a relevance score according to the semantic relevance calculation; Step 4: Combine the logical relationship score and the relevance score to calculate the logical coherence score of the entire text and obtain the analysis result; Step 5: Display the logical coherence score, logical relationship score and relevance score. Example

[0031] The embodiments of the present application also provide a computer-readable storage medium capable of implementing all the steps of the method for automatically analyzing the logical relationship of an article in the above embodiment, where the execution subject is a server or a client. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, all the steps of the method for automatically analyzing the logical relationship of an article in the above embodiment are implemented. For example, when the processor executes the computer program, the following steps are implemented: Step 1: Input the article text or paragraph text to be analyzed, and segment the input text into multiple sentences; Step 2: Analyze the grammatical structure and context information in the target sentence, identify the logical connection relationship between the target sentence and adjacent sentences, and obtain the logical relationship score; Step 3: Calculate the semantic relevance between the target sentence and the adjacent sentences according to the semantic contents of the target sentence and the adjacent sentences, and generate a relevance score according to the semantic relevance calculation; Step 4: Combine the logical relationship score and the relevance score to calculate the logical coherence score of the entire text and obtain the analysis result; Step 5: Display the logical coherence score, logical relationship score and relevance score.

[0032] The above-mentioned embodiments only express the preferred implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for ordinary technicians in this field, several modifications, improvements and substitutions can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application.

Claims

1. A method for automatically analyzing the logical relationship of an article, characterized in that: The following steps are involved: Step 1: Input the article text or paragraph text to be analyzed, and segment the input text into multiple sentences; Step 2: Analyze the grammatical structure and context information in the target sentence, identify the logical connection relationship between the target sentence and adjacent sentences, and obtain the logical relationship score; Step 3: Calculate the semantic relevance between the target sentence and the adjacent sentences according to the semantic contents of the target sentence and the adjacent sentences, and generate a relevance score according to the semantic relevance calculation; Step 4: Combine the logical relationship score and the relevance score to calculate the logical coherence score of the entire text and obtain the analysis result; Step 5: Display the logical coherence score, logical relationship score and relevance score.

2. The method for automatically analyzing the logical relationship of an article according to claim 1, characterized in that: The steps of identifying the logical relationship are as follows: Obtain excellent historical essays, count the logical relationships of the excellent historical essays, and obtain the logical connection pattern of the text under each theme; Preprocess the target sentence, including word segmentation, part-of-speech tagging, and named entity recognition; Determine whether the sentence has opinions and attitudes. If so, obtain the structure and annotation information of the target sentence and extract the logical relationship features of the sentence; The target sentence is scored according to the logical connection pattern and logical relationship features to obtain a logical relationship score.

3. The method for automatically analyzing the logical relationship of an article according to claim 2, characterized in that: The determination of whether a sentence has a viewpoint and an attitude is to determine whether a sentence has a strong viewpoint and an attitude, and the determination is made based on the classification of words, the position of the words, and whether the part where the words are located contains real meaning.

4. The method for automatically analyzing the logical relationship of an article according to claim 3, characterized in that: The specific process of judging whether a sentence has opinions and attitudes is as follows: Perform word segmentation on the target sentence, and traverse and judge the sentences after word segmentation; Determine whether the sentence contains strong attitude and opinion words. If so, judge that there is an attitude. If not, continue to the next round of judgment. Determine whether the sentence contains opinion nouns. If so, judge that there is an attitude. If not, proceed to the next round of judgment. Determine whether the sentence contains modal adverbs. If so, judge that it has an attitude. If not, proceed to the next round of judgment. Determine whether the sentence contains an emotional predicate verb. If so, it is judged to have an attitude. If not, continue to the next round of judgment. Determine whether the sentence contains polysex words. If so, judge that it has an attitude. If not, proceed to the next round of judgment. If there is no attitude word in the sentence, determine whether all the words in the sentence contain special phrases and structures. If so, it is judged that there is an attitude. If not, it is judged that there is no attitude.

5. The method for automatically analyzing the logical relationship of an article according to claim 4, characterized in that: Before performing word segmentation on a sentence, a data word library needs to be established. Multiple classification word libraries are established in the data word library, and multiple historical article texts and paragraph texts are obtained. The historical article texts and paragraph texts are segmented, and the different types of words after word segmentation are stored in the corresponding classification word library. The target sentence is traversed and judged through the words in the classification word library. The classification word library includes: a strong attitude and opinion word library, an opinion noun library, a modal adverb library, an emotional predicate verb library, a polysemic word library, and a special phrase library. After the traversal and judgment are completed, the words in the target sentence will be stored in the corresponding classification word library.

6. The method for automatically analyzing the logical relationship of an article according to claim 2, characterized in that: The calculation steps of the relevance score are as follows: Preprocess the two sentences to be compared, perform preliminary word segmentation and part-of-speech tagging, and obtain the sentence structure, which includes: keywords, part-of-speech information, and syntactic structure; Filter keywords according to their parts of speech, retain the key parts related to semantics, and remove irrelevant parts of the sentence; Perform keyword similarity matching on the target sentence based on the sentence structure. The matching priority is based on the logical relationship score from high to low to obtain the keyword matching score; The comprehensive matching rate of the target sentence is obtained by combining the matching score with the number and proportion of keywords, and the matching mode of the target sentence is determined according to the comprehensive matching rate; The matching score of each keyword is combined with the matching pattern to obtain the relevance score of the target sentence.

7. The method for automatically analyzing the logical relationship of an article according to claim 6, characterized in that: The steps of calculating the logical coherence score of the entire text are as follows: A comprehensive logical coherence evaluation model is established. The logical relationship scores and relevance scores are normalized according to the paragraph length. The logical weights and matching weights are calculated according to the logical connection patterns and matching patterns. The logical relationship scores and relevance scores are weighted according to the logical weights and matching weights to obtain a comprehensive logical coherence score.

8. An automatic analysis system for article logical relations, characterized in that: include: The text segmentation module is used to input the article text or paragraph text to be analyzed, and segment the input text into multiple sentences; The logic scoring module is used to analyze the grammatical structure and context information in the target sentence, identify the logical connection relationship between the target sentence and adjacent sentences, and obtain the logical relationship score; A relevance score module is used to calculate the semantic relevance between the target sentence and the adjacent sentences according to the semantic contents of the target sentence and the adjacent sentences, and generate a relevance score according to the semantic relevance calculation; A coherence score module is used to combine the logical relationship score and the relevance score to calculate the logical coherence score of the entire text and obtain the analysis result; The display module is used to display the logical coherence score, the logical relationship score and the relevance score.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method for automatically analyzing the logical relationship of an article as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for automatically analyzing the logical relationship of an article as described in any one of claims 1 to 7 are implemented.

Images