A method, system and storage medium for identifying intentions based on text semantic analysis

Abstract

The present invention discloses an intention recognition method, system and storage medium based on text semantic analysis, which relates to the technical field of text data processing, including S1, obtaining text data to be processed; S2, performing preprocessing based on the text data to be processed, and obtaining different sentence information in the text data; S3, performing feature extraction processing based on the different sentence information in the text data. The present invention first deletes non-language vocabulary words in the text data to be processed and retains qualified words. Secondly, the first processed text data is segmented according to the period type of the language punctuation mark to obtain different sentence information in the text data. Then, the correlation analysis of the different sentence information in the text data is performed to determine the weight information of different features. Finally, the weight information of different features is judged and processed to remove sentences that are irrelevant to the text intention. The above method can effectively reduce the computational complexity of text intention recognition and, at the same time, can also shorten the text intention recognition time.

Description

Technical Field

[0001] The present invention relates to the technical field of text data processing, and in particular to a method, system and storage medium for intent recognition based on text semantic analysis. Background Art

[0002] Text intent recognition uses natural language processing technology to automatically analyze and understand the intent or purpose expressed in text, enabling subsequent information processing or decision-making. The importance of text intent recognition: Text intent recognition has broad application value in fields such as human-computer interaction, intelligent question-and-answer (Q&A), information retrieval, and sentiment analysis. It helps machines understand human language, improves the efficiency and accuracy of human-computer interaction, enhances the quality and efficiency of intelligent Q&A systems, improves the accuracy of information retrieval results, and better understands user emotions and needs.

[0003] The text contains a large number of sentences, some of which have nothing to do with the text's intent and have no reference value for text intent recognition. If intent recognition is performed without deleting these sentences from the text, the computational complexity of text intent recognition will increase, and the time for text intent recognition will also be prolonged. Summary of the Invention

[0004] In order to solve the above technical problems, a method, system and storage medium for intent recognition based on text semantic analysis are provided. This technical solution solves the problem that the text proposed in the above background technology contains a large number of sentences, some of which have nothing to do with the intent of the text and have no reference value for the intent recognition of the text. If intent recognition is performed without deleting these sentences in the text, the computational complexity of text intent recognition will increase, and at the same time, the time for text intent recognition will be prolonged.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0006] An intent recognition method based on text semantic analysis, comprising:

[0007] S1. Obtain text data to be processed;

[0008] S2. Preprocess the text data to be processed to obtain information about different sentences in the text data;

[0009] S3, performing feature extraction processing based on different sentence information in the text data to determine different sentence feature information;

[0010] S4. Perform feature matching based on feature information of different sentences to obtain weight information of different features;

[0011] S5. Analyze and process the weight information of different features to determine the intention of the text.

[0012] Preferably, the step S2, performing preprocessing on the text data to be processed to obtain different sentence information in the text data, specifically comprises the following steps:

[0013] S21, randomly selecting words from the text data to be processed to obtain any number of text words;

[0014] S22, inputting any number of text words into the language database to perform language matching processing to determine the language type of the text words;

[0015] S23, performing data extraction processing on the language database based on the language type of the text vocabulary as a feature, and obtaining language-related data corresponding to the language type of the text vocabulary, wherein the language-related data includes a language vocabulary, language punctuation, and language grammar information;

[0016] S24. Preprocess the text data to be processed according to the language-related data to obtain information about different sentences in the text data.

[0017] Preferably, the step S24 of pre-processing the text data to be processed according to the language-related data to obtain information about different sentences in the text data specifically includes the following steps:

[0018] S241, uniquely numbering each word and punctuation mark in the text data to be processed, and obtaining words with different numbers and punctuation marks with different numbers;

[0019] S242, performing vocabulary matching processing on vocabulary with different numbers and language vocabulary lists, filtering out vocabulary in non-language vocabulary lists, and determining qualified vocabulary;

[0020] S243, resetting the qualified words according to their number information and punctuation marks with different numbers, to obtain first processed text data;

[0021] S244, performing data analysis and processing on the language punctuation marks to determine the period type of the language punctuation marks;

[0022] S245 , segmenting the first processed text data according to the period type of the language punctuation mark to obtain information of different sentences in the text data.

[0023] Preferably, the step S3, performing feature extraction processing based on different sentence information in the text data, and determining the feature information of different sentences specifically comprises the following steps:

[0024] S31, generating a unique virtual tag based on different sentence information in the text data, wherein the unique virtual tag internally records the length information of different sentences;

[0025] S32, performing part-of-speech analysis on different sentence information and language vocabularies in the text data to obtain part-of-speech information of the vocabulary;

[0026] S33. Perform feature analysis on the word part information and language grammar information of the vocabulary to determine feature information of different sentences.

[0027] Preferably, the step S4 of performing feature matching processing based on feature information of different sentences to obtain weight information of different features specifically includes the following steps:

[0028] S41, randomly selecting a piece of sentence feature information from different sentence feature information to obtain first sentence feature information;

[0029] S42, performing feature matching processing on the remaining different sentence feature information using the first sentence feature information to obtain sentence feature overlap;

[0030] S43, repeating steps S41-S42, performing feature matching processing on different sentence feature information to obtain sentence feature overlap;

[0031] S44, performing comparative analysis based on the overlap of the sentences to determine different sentences with relevance;

[0032] S45. Analyze and process different related sentences to obtain weight information of different features.

[0033] Preferably, the step S44 of performing comparative analysis based on the sentence overlap to determine different sentences with relevance specifically includes the following steps:

[0034] S441, performing comparison and judgment processing based on the sentence overlap and the set overlap threshold;

[0035] S442: If the sentence overlap is less than the set overlap threshold, the correlation between different sentence information in the two text data corresponding to the sentence overlap does not meet the standard;

[0036] S443. If the sentence overlap is greater than or equal to the set overlap threshold, the correlation between the different sentence information in the two text data corresponding to the sentence overlap meets the standard, and the different sentence information in the two text data corresponding to the sentence overlap is set as different sentences with correlation.

[0037] Preferably, the step S45 of analyzing and processing different related statements to obtain weight information of different features specifically includes the following steps:

[0038] S451, performing length extraction processing on unique virtual tags of different related sentences to obtain length information of the different related sentences;

[0039] S452: Calculate the length of the first processed text data to obtain the total length of the text data;

[0040] S453: Calculate and process the length information of different related sentences and the total length information of the text data to obtain weight information of different features.

[0041] Preferably, the step S5 of analyzing and processing the weight information of different features to determine the intention of the text specifically includes the following steps:

[0042] S51, comparing and judging the weight information of different features and the set weight threshold;

[0043] S52: If the weight information of different features is greater than or equal to the set weight threshold, the weight information of different features meets the standard, and the intention of different sentences in the text data corresponding to the weight information of different features is identified to determine the intention of the text;

[0044] S53. If the weight information of different features is less than the set weight threshold, the weight information of different features does not meet the standard, and intention recognition is not performed on different sentence information in the text data corresponding to the weight information of different features.

[0045] Furthermore, a text semantic analysis-based intention recognition system is proposed, which is used to implement the above-mentioned text semantic analysis-based intention recognition method, including:

[0046] An intention recognition module is used to control each module to perform text preprocessing, feature extraction, feature matching, and weight analysis on the text data to be processed, determine the intention of the text, and control data transmission and information exchange between each module;

[0047] A database system, wherein the database system is used to store text data to be processed;

[0048] A language library, which is used to store various language vocabularies, language punctuation marks, and language grammar information;

[0049] A text data preprocessing module, which is used to screen the text data to be processed, delete non-language words, and determine different sentence information in the text data;

[0050] A label generation module, wherein the label generation module generates unique virtual labels according to the length information of different sentences;

[0051] A feature extraction module is used to perform part-of-speech analysis and feature extraction on different sentence information in the text data to determine feature information of different sentences;

[0052] A feature matching module is used to perform feature matching between feature information of different sentences, determine the overlap of sentence features, and obtain different sentences with relevance;

[0053] A weight calculation module is used to calculate the length information of different related sentences and the total length information of the text data to obtain weight information of different features;

[0054] A text intent recognition module is used to judge and process the weight information of different features, determine the weight information of different features that meet the standards, and then perform intent recognition on different sentence information in the text data corresponding to the weight information of different features that meet the standards to determine the intent of the text.

[0055] Furthermore, a storage medium is proposed, on which a computer program is stored. When the computer program is called and run, it executes the above-mentioned intention recognition method based on text semantic analysis.

[0056] Compared with the prior art, the present invention provides a method, system, and storage medium for intent recognition based on text semantic analysis, which has the following beneficial effects:

[0057] The present invention first deletes non-language vocabulary words in the text data to be processed and retains qualified vocabulary. Secondly, the first processed text data is segmented according to the period type of the language punctuation mark to obtain different sentence information in the text data. Then, the correlation analysis of the different sentence information in the text data is performed to determine the weight information of different features. Finally, the weight information of different features is judged and processed to remove sentences that are not related to the text intent. The above method can effectively reduce the computational amount of text intent recognition and, at the same time, shorten the text intent recognition time. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is a flow chart of steps S1-S5 in the method for intention recognition based on text semantic analysis proposed by the present invention;

[0059] Figure 2 This is a flowchart of steps S21-S24 in the method for intention recognition based on text semantic analysis proposed by the present invention;

[0060] Figure 3 This is a flow chart of steps S241-S245 in the method for intention recognition based on text semantic analysis proposed by the present invention;

[0061] Figure 4This is a flowchart of steps S31-S33 in the method for intention recognition based on text semantic analysis proposed by the present invention;

[0062] Figure 5 This is a flow chart of steps S41-S45 in the method for intention recognition based on text semantic analysis proposed by the present invention;

[0063] Figure 6 This is a flowchart of steps S441-S443 in the method for intention recognition based on text semantic analysis proposed by the present invention;

[0064] Figure 7 Schematic diagram of the flow of steps S451-S453 in the method for intention recognition based on text semantic analysis proposed by the present invention;

[0065] Figure 8 This is a flowchart of steps S51-S53 in the method for intention recognition based on text semantic analysis proposed by the present invention;

[0066] Figure 9 This is a structural block diagram of an intent recognition system based on text semantic analysis proposed by the present invention. DETAILED DESCRIPTION

[0067] The following description is intended to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are merely examples, and those skilled in the art may conceive of other obvious variations.

[0068] Reference Figure 1 As shown, an intent recognition method based on text semantic analysis includes:

[0069] S1. Obtain text data to be processed;

[0070] S2. Preprocess the text data to be processed to obtain information about different sentences in the text data;

[0071] S3, performing feature extraction processing based on different sentence information in the text data to determine different sentence feature information;

[0072] S4. Perform feature matching based on feature information of different sentences to obtain weight information of different features;

[0073] S5. Analyze and process the weight information of different features to determine the intention of the text;

[0074] Those skilled in the art can understand that a text is composed of a number of groups of sentences arranged and combined. Some sentences in the text are valid sentences, while others are invalid sentences. When it is necessary to perform intent analysis on the text, if the invalid sentences are not deleted before performing the intent analysis, the computational complexity of the intent recognition module will be increased. At the same time, due to too many sentences, the time for text intent recognition will be extended. Therefore, it is necessary to delete the invalid sentences in the text to reduce the computational complexity of the intent recognition module and shorten the text intent recognition time.

[0075] Reference Figure 2 As shown, S2, preprocessing the text data to be processed, obtaining different sentence information in the text data specifically includes the following steps:

[0076] S21, randomly selecting words from the text data to be processed to obtain any number of text words;

[0077] Some languages may have some of the same words. If only one word is extracted, it may cause language judgment errors. Therefore, a random number function is used to randomly select words from the text data to be processed to avoid misjudgment. For example, Japanese and Chinese have some of the same words.

[0078] S22, inputting any number of text words into the language database to perform language matching processing to determine the language type of the text words;

[0079] S23, performing data extraction processing on the language database based on the language type of the text vocabulary as a feature, and obtaining language-related data corresponding to the language type of the text vocabulary, wherein the language-related data includes a language vocabulary, language punctuation, and language grammar information;

[0080] S24, preprocessing the text data to be processed based on the language-related data to obtain information about different sentences in the text data;

[0081] In this embodiment, the language of the text stored in the database system may be diverse, so it is necessary to determine the vocabulary types in the text data to be processed. Because different languages may have different composition methods and grammars, the text data to be processed is subjected to language analysis to determine its language type, and invalid sentences in the text data to be processed are deleted in combination with its language vocabulary, language punctuation and language grammar information to reduce the computational complexity of the intent recognition module.

[0082] Reference Figure 3 As shown, S24, pre-processing the text data to be processed according to the language-related data, and obtaining different sentence information in the text data specifically includes the following steps:

[0083] S241, uniquely numbering each word and punctuation mark in the text data to be processed, and obtaining words with different numbers and punctuation marks with different numbers;

[0084] S242, performing vocabulary matching processing on vocabulary with different numbers and language vocabulary lists, filtering out vocabulary in non-language vocabulary lists, and determining qualified vocabulary;

[0085] S243, resetting the qualified words according to their number information and punctuation marks with different numbers, to obtain first processed text data;

[0086] S244, performing data analysis and processing on the language punctuation marks to determine the period type of the language punctuation marks;

[0087] S245, segmenting the first processed text data according to the period type of the language punctuation mark to obtain different sentence information in the text data;

[0088] In this embodiment, the text data to be processed stored in the database system may be a hand-drawn draft, and there may be some corrections in the hand-drawn draft. Therefore, by comparing the words in the text data to be processed with the language vocabulary, those positions with corrections are deleted. However, extracting the words in the text data to be processed for comparison may cause misplacement when the words are subsequently reset. Therefore, each word and punctuation mark in the text data to be processed is uniquely numbered. After deleting those positions with corrections, they can be reset according to the unique number of each word and punctuation mark, so that the overall meaning of the text remains unchanged. At the same time, it also avoids the subsequent intention recognition module from processing those positions with corrections, reducing the computational complexity of the intention recognition module. In addition, the punctuation marks are also uniquely numbered because punctuation marks are the signs of sentence segmentation. If the punctuation marks are confusing, the semantics of the text data to be processed will be unclear, and then text intention recognition errors will be caused.

[0089] Reference Figure 4 As shown, S3, performing feature extraction processing based on different sentence information in the text data, and determining the feature information of different sentences specifically includes the following steps:

[0090] S31, generating a unique virtual tag based on different sentence information in the text data, wherein the unique virtual tag internally records the length information of different sentences;

[0091] It is understandable that different sentences have different lengths, so sentences of different lengths will account for different proportions in the text data to be processed. Therefore, the purpose of generating unique virtual labels is to avoid recalculating the lengths of different sentences when performing feature analysis on them later.

[0092] S32, performing part-of-speech analysis on different sentence information and language vocabularies in the text data to obtain part-of-speech information of the vocabulary;

[0093] S33, performing feature analysis on the word part information and language grammar information of the vocabulary to determine feature information of different sentences;

[0094] The bag-of-words model is used to classify the words in different sentence information in the text data. The word model treats each word in the text as a separate feature.

[0095] In this embodiment, different sentences have different vocabulary compositions and may also have different parts of speech. For example, "I have been to Nanjing" and "I want to go to Nanjing" have different parts of speech, resulting in different text intentions. Therefore, feature extraction is performed on different sentences through part of speech information and language grammar information to determine feature information of different sentences.

[0096] Reference Figure 5 As shown, S4, performing feature matching processing based on different sentence feature information to obtain weight information of different features specifically includes the following steps:

[0097] S41, randomly selecting a piece of sentence feature information from different sentence feature information to obtain first sentence feature information;

[0098] S42, performing feature matching processing on the remaining different sentence feature information using the first sentence feature information to obtain sentence feature overlap;

[0099] Perform feature matching processing on different sentence feature information through neural network algorithm to determine the overlap of sentence features;

[0100] S43, repeating steps S41-S42, performing feature matching processing on different sentence feature information to obtain sentence feature overlap;

[0101] S44, performing comparative analysis based on the overlap of the sentences to determine different sentences with relevance;

[0102] S45. Analyze and process different related sentences to obtain weight information of different features;

[0103] In this embodiment, in a text, key features and sub-key features may appear multiple times, but the types of appearance may be different. They may appear in one form in one sentence and in another form in another sentence. For example, "I want to go to Nanjing" and "Can you recommend famous attractions in Nanjing to me?" The above two sentences both mention Nanjing, but Nanjing appears in different situations. Therefore, by matching the feature information of different sentences, the feature overlap between different sentences is determined.

[0104] Reference Figure 6 As shown, S44, performing comparative analysis based on the sentence overlap to determine different sentences with relevance specifically includes the following steps:

[0105] S441, performing comparison and judgment processing based on the sentence overlap and the set overlap threshold;

[0106] S442: If the sentence overlap is less than the set overlap threshold, the correlation between different sentence information in the two text data corresponding to the sentence overlap does not meet the standard;

[0107] S443: If the sentence overlap is greater than or equal to the set overlap threshold, then the correlation between the different sentence information in the two text data corresponding to the sentence overlap meets the standard, and the different sentence information in the two text data corresponding to the sentence overlap is set as different sentences with correlation;

[0108] In this embodiment, if the feature overlap between different statements is small, it means that the features between the two statements are different, one may be a key feature and the other may be an irrelevant feature. Therefore, the statement overlap and the set overlap threshold are judged to determine whether the correlation between the two different statements meets the standard.

[0109] Reference Figure 7 As shown, S45, analyzing and processing different related sentences to obtain weight information of different features specifically includes the following steps:

[0110] S451, performing length extraction processing on unique virtual tags of different related sentences to obtain length information of the different related sentences;

[0111] S452: Calculate the length of the first processed text data to obtain the total length of the text data;

[0112] S453: Calculate and process the length information of different related sentences and the total length information of the text data to obtain weight information of different features.

[0113] Reference Figure 8 As shown, S5, analyzing and processing the weight information of different features to determine the intention of the text specifically includes the following steps:

[0114] S51, comparing and judging the weight information of different features and the set weight threshold;

[0115] S52: If the weight information of different features is greater than or equal to the set weight threshold, the weight information of different features meets the standard, and the intention of different sentences in the text data corresponding to the weight information of different features is identified to determine the intention of the text;

[0116] S53: If the weight information of different features is less than the set weight threshold, the weight information of different features does not meet the standard, and the intention recognition is not performed on the different sentence information in the text data corresponding to the weight information of different features;

[0117] In this embodiment, if there are irrelevant sentences in the text data to be processed, the weight of the irrelevant sentences in the text data to be processed will be very small, because the irrelevant sentences may be mentioned casually in the text data to be processed, with only one or two sentences, and the correlation between the irrelevant sentences and the intention in the text data to be processed will also be very small. Therefore, by comparing and judging the weight information of different features and setting the weight threshold, the irrelevant sentences are deleted, which reduces the computational complexity of the intention recognition module and, at the same time, shortens the intention recognition time of the text data to be processed.

[0118] Reference Figure 9 As shown, an intention recognition system based on text semantic analysis is used to implement the above-mentioned intention recognition method based on text semantic analysis, including:

[0119] An intention recognition module is used to control each module to perform text preprocessing, feature extraction, feature matching, and weight analysis on the text data to be processed, determine the intention of the text, and control data transmission and information exchange between each module;

[0120] A database system, wherein the database system is used to store text data to be processed;

[0121] A language library, which is used to store various language vocabularies, language punctuation marks, and language grammar information;

[0122] A text data preprocessing module, which is used to screen the text data to be processed, delete non-language words, and determine different sentence information in the text data;

[0123] A label generation module, wherein the label generation module generates unique virtual labels according to the length information of different sentences;

[0124] A feature extraction module is used to perform part-of-speech analysis and feature extraction on different sentence information in the text data to determine feature information of different sentences;

[0125] A feature matching module is used to perform feature matching between feature information of different sentences, determine the overlap of sentence features, and obtain different sentences with relevance;

[0126] A weight calculation module is used to calculate the length information of different related sentences and the total length information of the text data to obtain weight information of different features;

[0127] A text intent recognition module is used to judge and process the weight information of different features, determine the weight information of different features that meet the standards, and then perform intent recognition on different sentence information in the text data corresponding to the weight information of different features that meet the standards to determine the intent of the text.

[0128] Furthermore, a storage medium is proposed, on which a computer program is stored. When the computer program is called and run, it executes an intention recognition method based on text semantic analysis as described above, wherein the storage medium can be a magnetic medium, such as a floppy disk, a hard disk, or a tape; an optical medium, such as a DVD; or a semiconductor medium, such as a solid-state drive (SSD).

[0129] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions merely illustrate the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for identifying intentions based on text semantic analysis, characterized in that: include: S1. Obtain text data to be processed; S2. Preprocess the text data to be processed to obtain information about different sentences in the text data; S3, performing feature extraction processing based on different sentence information in the text data to determine different sentence feature information; S4. Perform feature matching based on different sentence feature information to obtain weight information of different features; S5. Analyze and process the weight information of different features to determine the intention of the text; The step S2, performing preprocessing on the text data to be processed to obtain different sentence information in the text data, specifically includes the following steps: S21, randomly selecting words from the text data to be processed to obtain any number of text words; S22, inputting any number of text words into the language database to perform language matching processing to determine the language type of the text words; S23, performing data extraction processing on the language database based on the language type of the text vocabulary as a feature, and obtaining language-related data corresponding to the language type of the text vocabulary, wherein the language-related data includes a language vocabulary, language punctuation, and language grammar information; S24, pre-processing the text data to be processed according to the language-related data, and obtaining different sentence information in the text data specifically includes the following steps: S241, uniquely numbering each word and punctuation mark in the text data to be processed, and obtaining words with different numbers and punctuation marks with different numbers; S242, performing vocabulary matching processing on vocabulary with different numbers and language vocabulary lists, filtering out vocabulary in non-language vocabulary lists, and determining qualified vocabulary; S243, resetting the qualified words according to their number information and punctuation marks with different numbers, to obtain first processed text data; S244, performing data analysis and processing on the language punctuation marks to determine the period type of the language punctuation marks; S245 , segmenting the first processed text data according to the period type of the language punctuation mark to obtain information of different sentences in the text data.

2. The method for identifying intentions based on text semantic analysis according to claim 1, characterized in that: The step S3, performing feature extraction processing based on different sentence information in the text data, and determining the feature information of different sentences specifically includes the following steps: S31, generating a unique virtual tag based on different sentence information in the text data, wherein the unique virtual tag internally records the length information of different sentences; S32, performing part-of-speech analysis on different sentence information and language vocabularies in the text data to obtain part-of-speech information of the vocabulary; S33. Perform feature analysis on the word part information and language grammar information of the vocabulary to determine feature information of different sentences.

3. The method for identifying intentions based on text semantic analysis according to claim 1, wherein: The step S4, performing feature matching based on different sentence feature information to obtain weight information of different features, specifically includes the following steps: S41, randomly selecting a piece of sentence feature information from different sentence feature information to obtain first sentence feature information; S42, performing feature matching processing on the remaining different sentence feature information using the first sentence feature information to obtain sentence feature overlap; S43, repeating steps S41-S42, performing feature matching processing on different sentence feature information to obtain sentence feature overlap; S44, performing comparative analysis based on the overlap of the sentences to determine different sentences with relevance; S45. Analyze and process different related sentences to obtain weight information of different features.

4. The method for identifying intentions based on text semantic analysis according to claim 3, wherein: The step S44 of performing comparative analysis based on the sentence overlap to determine different sentences with relevance specifically includes the following steps: S441, performing comparison and judgment processing based on the sentence overlap and the set overlap threshold; S442: If the sentence overlap is less than the set overlap threshold, the correlation between different sentence information in the two text data corresponding to the sentence overlap does not meet the standard; S443. If the sentence overlap is greater than or equal to the set overlap threshold, the correlation between the different sentence information in the two text data corresponding to the sentence overlap meets the standard, and the different sentence information in the two text data corresponding to the sentence overlap is set as different sentences with correlation.

5. The method for identifying intentions based on text semantic analysis according to claim 3, wherein: The step S45 of analyzing and processing different related statements to obtain weight information of different features specifically includes the following steps: S451, performing length extraction processing on unique virtual tags of different related sentences to obtain length information of the different related sentences; S452: Calculate the length of the first processed text data to obtain the total length of the text data; S453: Calculate and process the length information of different related sentences and the total length information of the text data to obtain weight information of different features.

6. The method for identifying intentions based on text semantic analysis according to claim 1, characterized in that: The step S5 of analyzing and processing the weight information of different features to determine the intention of the text specifically includes the following steps: S51, comparing and judging the weight information of different features and the set weight threshold; S52: If the weight information of different features is greater than or equal to the set weight threshold, the weight information of different features meets the standard, and the intention of different sentences in the text data corresponding to the weight information of different features is identified to determine the intention of the text; S53. If the weight information of different features is less than the set weight threshold, the weight information of different features does not meet the standard, and intention recognition is not performed on different sentence information in the text data corresponding to the weight information of different features.

7. An intention recognition system based on text semantic analysis, used to implement the intention recognition method based on text semantic analysis according to any one of claims 1 to 6, characterized in that: include: An intention recognition module is used to control each module to perform text preprocessing, feature extraction, feature matching, and weight analysis on the text data to be processed, determine the intention of the text, and control data transmission and information exchange between each module; A database system, wherein the database system is used to store text data to be processed; A language library, which is used to store various language vocabularies, language punctuation marks, and language grammar information; A text data preprocessing module, which is used to screen the text data to be processed, delete non-language words, and determine different sentence information in the text data; A label generation module, wherein the label generation module generates unique virtual labels according to the length information of different sentences; A feature extraction module is used to perform part-of-speech analysis and feature extraction on different sentence information in the text data to determine feature information of different sentences; A feature matching module is used to perform feature matching between feature information of different sentences, determine the overlap of sentence features, and obtain different sentences with relevance; A weight calculation module is used to calculate the length information of different related sentences and the total length information of the text data to obtain weight information of different features; A text intent recognition module is used to judge and process the weight information of different features, determine the weight information of different features that meet the standards, and then perform intent recognition on different sentence information in the text data corresponding to the weight information of different features that meet the standards to determine the intent of the text.

8. A storage medium, characterized in that: A computer program is stored thereon, and when the computer program is called and run, it executes the intention recognition method based on text semantic analysis as described in any one of claims 1 to 6.

Images