Artificial Intelligence-Based Contract Content Query Method and System

Abstract

This application relates to the field of data processing technology and discloses a method and system for querying contract content based on artificial intelligence. The method includes: splitting the contract text into multiple text blocks; performing semantic analysis on each text block to extract keywords, including a first keyword and a second keyword; constructing a contract information structure diagram based on the keywords; obtaining word vectors for each keyword based on a pre-trained first model; calculating the proximity between every two keywords in the text block based on the word vectors; improving the contract information structure diagram based on the proximity and creating a first association table; obtaining query keywords based on real-time query statements; finding keywords with the same frequency as the query keywords based on historical query information; extracting first information and second information from the contract information structure diagram based on the query keywords, keywords with the same frequency, and the first association table; and displaying the query results based on the first information and second information. This invention can improve the efficiency of contract content querying.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method and system for querying contract content based on artificial intelligence. Background Technology

[0002] In business activities, contracts are crucial legal documents, and the retrieval and management of their contents are essential for business operations. Traditional contract retrieval methods primarily rely on manual searching, which is inefficient and prone to errors. With the development of artificial intelligence technology, technologies such as natural language processing and knowledge graphs have provided new solutions for contract content retrieval.

[0003] A similar prior art patent application, CN113505303A, discloses a method and apparatus for querying contract information, comprising: obtaining contract information to be queried input by a user; the contract information to be queried includes keywords; querying whether the keywords support searching the contract information to be queried in an ES browser; if the keywords support searching the contract information to be queried in an ES browser, then searching for a contract identity identifier containing the contract information to be queried in an ES browser, and querying the contract information corresponding to the contract identity identifier in a database based on the contract identity identifier; if the keywords do not support searching the contract information to be queried in an ES browser, then searching for the contract information to be queried in a database, and querying the contract information containing the contract information to be queried based on the contract information to be queried.

[0004] Similar existing technologies include Chinese patent application CN113033197A, which provides a method and apparatus for querying construction contract regulations, including: collecting construction contract regulations and digitizing them to establish a construction contract regulation database; performing text segmentation and stop word removal on the construction contract regulations based on natural language processing technology, and then calculating feature words through a word frequency inverse text algorithm; performing synonym expansion query on the feature words through a self-built lexicon of commonly used terms in construction contract regulations and a continuous bag-of-words model; calculating the similarity of contract regulations based on a vector space model and an improved cosine function method to obtain the corresponding legal provisions in the "Conditions of Construction Contract"; and integrating the entire database and query system into a local server or smart device.

[0005] However, most existing contract query systems can only achieve simple text matching and cannot accurately understand the semantics and logical relationships of contract content. The accuracy, relevance, and efficiency of query results need to be improved.

[0006] Therefore, the present invention provides a method and system for querying contract content based on artificial intelligence. Summary of the Invention

[0007] This application provides a method and system for querying contract content based on artificial intelligence, which can be used to efficiently and accurately query contract content.

[0008] Firstly, this application provides a method for querying contract content based on artificial intelligence, the method comprising:

[0009] The contract text is split into multiple text blocks, and semantic parsing is performed on each text block to extract keywords. The keywords include the first keyword and the second keyword. A contract information structure diagram is constructed based on the keywords.

[0010] The word vectors of each keyword are obtained based on the pre-trained first model. The proximity between each pair of keywords in the text block is calculated based on the word vectors. The contract information structure graph is improved based on the proximity and a first association table is created.

[0011] Store historical query information, obtain query keywords based on real-time query statements, and find keywords with the same frequency that appear at the same time as the query keywords based on historical query information;

[0012] Based on query keywords, similar keywords, and the first association table, the first and second information are extracted from the contract information structure diagram, and the query results are displayed based on the first and second information.

[0013] In conjunction with the first aspect, in the first implementation of the first aspect of this application, a contract information structure diagram is constructed based on keywords, including:

[0014] The first keyword is used as the key point in the contract information structure diagram. It is determined whether there is a relationship between any two first keywords. If so, the first connecting line is added to the two key points corresponding to the two first keywords that are related, and the relationship between the two first keywords is marked.

[0015] In conjunction with the first aspect, in the second implementation of the first aspect of this application, the contract information structure diagram is improved based on proximity, including:

[0016] First keywords with a proximity greater than a preset first threshold are combined to generate a first keyword group. A keyword is selected from the first keyword group as the target keyword. The first keyword in the contract information structure diagram that belongs to the same first keyword group as the target keyword is replaced with the target keyword.

[0017] The second keyword group is generated by combining the second keywords whose proximity is greater than the first threshold. The third keyword is generated based on the second keyword group. The corresponding key point is added to the contract information structure diagram based on the third keyword. A second connecting line is also added between the key point corresponding to the third keyword and the keywords in the second keyword group.

[0018] In conjunction with the first aspect, the third implementation of the first aspect of this application improves the contract information structure diagram based on proximity, including:

[0019] Obtain the number of first connecting lines connected to each key point, identify the first keyword corresponding to the key point whose number of lines is greater than a preset second threshold as a feature keyword, and record all feature keywords and their positions in the contract information structure diagram.

[0020] A standard thesaurus is pre-created, which stores standard words with high query frequency. For each keyword in the contract information structure diagram, the proximity between it and the standard words in the standard thesaurus is calculated. The first standard word with a proximity greater than a first threshold is obtained. The first standard word with the largest proximity is obtained as the second standard word. The corresponding keyword is replaced with the second standard word. The second standard word and the keyword are also associated and saved to the first association table.

[0021] In conjunction with the first aspect, in the fourth implementation of the first aspect of this application, the search for and query keywords that appear simultaneously based on historical query information includes:

[0022] All keywords in the historical query information are called fourth keywords. The fourth keywords are combined with the query keywords in pairs to generate multiple keyword pairs.

[0023] The historical query information is divided into multiple different sub-information sets according to time periods. The historical keyword groups in each sub-information set are extracted. For each keyword pair, the first occurrence in each historical keyword group is counted, as well as the second and third occurrences.

[0024] Compare the second and third counts, find the maximum value, add all the first counts together to get the first result value, divide the first result value by the maximum value to get the second result value, find the fourth keyword corresponding to the second result value that is greater than the second threshold, and use the fourth keyword as the same frequency keyword of the corresponding query keyword.

[0025] In conjunction with the first aspect, in the fifth implementation of the first aspect of this application, the first information and the second information are extracted from the contract information structure diagram, including:

[0026] Search and query the standard words that match the keywords from the first association table. If the standard words that match are found, obtain the candidate keywords based on the first association table.

[0027] Determine whether there are characteristic keywords that match the candidate keywords. If there are, obtain the position of the characteristic keywords in the contract information structure diagram and the corresponding key points. If not, find the key points that match the candidate keywords from the contract information structure diagram.

[0028] Based on the connecting lines of key points, related key points are found, and the found key points, related key points and corresponding connecting lines are used as the first information.

[0029] In conjunction with the first aspect, in the sixth implementation of the first aspect of this application, when no matching standard word can be found, the method includes: searching for a standard word that matches the same frequency keyword from the first association table; and when a matching standard word is found, obtaining candidate keywords based on the first association table.

[0030] Determine whether there are characteristic keywords that match the same frequency keywords. If there are, obtain the position of the characteristic keywords in the contract information structure diagram and the corresponding key points. If not, search for key points that match the candidate keywords in the contract information structure diagram.

[0031] Based on the connecting lines of key points, related key points are found, and the found key points, related key points, and corresponding connecting lines are used as secondary information.

[0032] In conjunction with the first aspect, in the seventh implementation of the first aspect of this application, the query results are displayed based on the first information and the second information, including:

[0033] Extract duplicate information from the first and second information, place the duplicate information in the first position of the query results, and sort and display other information based on its proximity to the query keywords.

[0034] Secondly, this application provides an artificial intelligence-based contract content query system, the system comprising:

[0035] The structure diagram generation unit is used to split the contract text into multiple text blocks, perform semantic parsing on each text block to extract keywords, including the first keyword and the second keyword, and construct a contract information structure diagram based on the keywords;

[0036] The structure graph update unit is used to obtain the word vector of each keyword based on the pre-trained first model, calculate the proximity between every two keywords in the text block based on the word vector, improve the contract information structure graph based on the proximity, and create the first association table.

[0037] The homonym search unit is used to store historical query information, obtain query keywords based on real-time query statements, and search for homonyms that appear at the same time as the query keywords based on historical query information.

[0038] The result query unit is used to extract first information and second information from the contract information structure diagram based on query keywords, homonymous keywords and the first association table, and to display the query results based on the first information and second information.

[0039] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0040] The technical solution provided in this application splits the contract text into multiple text blocks, performs semantic parsing on each text block to extract keywords, and constructs a contract information structure diagram based on the keywords. This logically categorizes and associates the contract content, enabling users to quickly locate key information during queries, significantly reducing the search scope and improving query speed. By improving the contract information structure diagram and creating a first association table, the structure of the contract structure diagram is optimized to more accurately express semantic relationships. By using keywords with similar frequencies, users can find more comprehensive content, improving query efficiency. The query results are displayed based on first and second information, allowing users to find not only content directly related to the query keywords but also other related information. Furthermore, the first and second information are sorted based on proximity, helping users obtain more comprehensive, accurate, and intuitive query results. Attached Figure Description

[0041] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a schematic diagram of an embodiment of the contract content query method based on artificial intelligence in this application.

[0043] Figure 2 This is a contract information structure diagram generated based on a text block of a lease contract in an embodiment of this application;

[0044] Figure 3 This is the updated contract information structure diagram in the embodiments of this application;

[0045] Figure 4 This is a schematic diagram of one embodiment of the contract content query system based on artificial intelligence in this application. Detailed Implementation

[0046] This application provides a method and system for querying contract content based on artificial intelligence. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0047] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the contract content query method based on artificial intelligence in this application includes:

[0048] Step S1: Divide the contract text into multiple text blocks, perform semantic parsing on each text block to extract keywords, including the first keyword and the second keyword, and construct a contract information structure diagram based on the keywords.

[0049] Specifically, to improve the intelligence level of contract management and provide users with efficient and accurate query services, the contract text is first divided into multiple text blocks. For example, a section of the contract text can be considered as a text block. Then, semantic parsing is performed on each text block to extract keywords. For example, natural language processing technology can be used to extract keywords from the text blocks. Keywords include primary keywords and secondary keywords. Primary keywords are nouns, including the two parties to the contract, obligations, rights, and clause names, etc. Secondary keywords are verbs, including words that can express the relationship between the two primary keywords, such as "payment" and "providing a certain service." Then, a contract information structure diagram is constructed based on the keywords. The specific generation method will be explained in detail later. The generated contract information structure diagram is stored in a graph database. By structuring the contract content into key points and lines, and logically classifying and associating the contract content, the contract information structure diagram can quickly locate the key information during querying, greatly reducing the search scope and improving query speed.

[0050] Step S2: Obtain the word vector of each keyword based on the pre-trained first model, calculate the proximity between every two keywords in the text block based on the word vector, improve the contract information structure diagram based on the proximity, and create the first association table.

[0051] Specifically, to optimize the structure of the contract information graph and enable it to more accurately express semantic relationships, we first obtain word vectors for each keyword based on a pre-trained first model, such as the Word2Vec model. Then, based on the word vectors, we use a similarity algorithm, such as cosine similarity, to calculate the proximity between every two keywords in the text block. After that, we optimize the contract information graph based on the proximity. The specific optimization methods will be explained in detail later. While optimizing the contract information graph, we also create a first association table. The first association table stores the keywords in the contract content and their corresponding standard words, as well as the positions of the keywords in the contract information graph. The first association table can assist in subsequent queries of the contract content.

[0052] Step S3: Store historical query information, obtain query keywords based on real-time query statements, and find keywords with the same frequency that appear at the same time as the query keywords based on historical query information.

[0053] Specifically, to enhance the convenience and accuracy of user information retrieval, the system associates and saves user-inputted queries with query times to generate historical query information. When users subsequently input real-time queries, natural language processing (NLP) technology is used to extract query keywords. Then, based on historical query information, co-frequency keywords that appear simultaneously with the query keywords are obtained. Co-frequency keywords are those that are frequently entered by users within a preset time period, similar to the query keywords. For example, when a user searches for breach of contract liability, they might also input compensation amounts or breach notice periods within a short period before and after the query. These keywords are co-frequency keywords. The specific methods for obtaining co-frequency keywords will be explained in detail later. After obtaining co-frequency keywords, more comprehensive content can be retrieved for users based on the query keywords and co-frequency keywords, improving the efficiency of user queries.

[0054] Step S4: Extract first information and second information from the contract information structure diagram based on query keywords, homonymous keywords and the first association table, and display the query results based on the first information and second information.

[0055] Specifically, the first information refers to the content information found in the contract information structure diagram based on the query keywords that is directly or indirectly related to the query keywords. The second information refers to the content information found in the contract information structure diagram based on keywords with similar frequency that is directly or indirectly related to the query keywords. Subsequently, the first and second information are displayed based on their proximity to the query keywords. This provides users with comprehensive query results while making it convenient for them to view the results intuitively.

[0056] In one specific embodiment, constructing a contract information structure diagram based on keywords also involves the following steps:

[0057] The first keyword is used as the key point in the contract information structure diagram. It is determined whether there is a relationship between any two first keywords. If so, the first connecting line is added to the two key points corresponding to the two first keywords that are related, and the relationship between the two first keywords is marked.

[0058] Specifically, to improve the efficiency of contract content retrieval, a contract information structure diagram is constructed. First, the primary keyword is used as a key point in the diagram. Then, based on the correlation between primary keywords, a first connecting line is added between key points corresponding to two related primary keywords. The correlation between two key points (corresponding to two primary keywords) is also marked on this first connecting line, such as... Figure 2 The diagram shown is a contract information structure diagram generated from a text block of a lease contract. The generated contract information structure diagram is converted into graph data and stored in a graph database. The structured form of the graph data facilitates subsequent querying and updating.

[0059] In one specific embodiment, improving the contract information structure diagram based on proximity includes the following steps:

[0060] First keywords with a proximity greater than a preset first threshold are combined to generate a first keyword group. A keyword is selected from the first keyword group as the target keyword. The first keyword in the contract information structure diagram that belongs to the same first keyword group as the target keyword is replaced with the target keyword.

[0061] The second keyword group is generated by combining the second keywords whose proximity is greater than the first threshold. The third keyword is generated based on the second keyword group. The corresponding key point is added to the contract information structure diagram based on the third keyword. A second connecting line is also added between the key point corresponding to the third keyword and the keywords in the second keyword group.

[0062] Specifically, in order to reduce redundancy in the contract information structure diagram and improve its simplicity, first keywords with a similarity greater than a preset first threshold are combined to generate a first keyword group. The first keyword group contains multiple semantically similar first keywords. One first keyword is selected from the first keyword group as the target keyword. For example, if a group of first keywords contains "house" and "real estate", then one of them, "real estate", can be selected as the target keyword, and all "houses" in the contract information structure diagram can be replaced with "real estate".

[0063] To enrich the content of the contract information structure diagram and enable it to more accurately express the semantic relationships between words, second keywords with a proximity greater than a first threshold are combined to generate second keyword groups. For example, one set of second keyword groups might be "payment" and "delivery." Based on these second keyword groups, third keywords are generated. The third keyword is a summary term that can summarize the second keyword group. For example, the generated third keyword might be "payment." Key points corresponding to the third keyword are added to the contract information structure diagram, and second connecting lines are added between the keywords in the second keyword group and the third keyword. The second connecting lines represent that there is a summary relationship between the third keyword and the second keyword.

[0064] The above methods enable the contract information structure diagram to more comprehensively express the information in the contract, avoid missing important semantic relationships, and quickly find relevant key points through third keywords during subsequent queries.

[0065] In one specific embodiment, the process of improving the contract information structure graph based on proximity also includes the following steps:

[0066] Obtain the number of first connecting lines connected to each key point, identify the first keyword corresponding to the key point whose number of lines is greater than a preset second threshold as a feature keyword, and record all feature keywords and their positions in the contract information structure diagram.

[0067] A standard thesaurus is pre-created, which stores standard words with high query frequency. For each keyword in the contract information structure diagram, the proximity between it and the standard words in the standard thesaurus is calculated. The first standard word with a proximity greater than a first threshold is obtained. The first standard word with the largest proximity is obtained as the second standard word. The corresponding keyword is replaced with the second standard word. The second standard word and the keyword are also associated and saved to the first association table.

[0068] Specifically, in order to further improve the contract information structure diagram, the number of first connecting lines connected to each key point in the contract information structure diagram is obtained. The more lines there are, the more other first keywords are related to the first keyword corresponding to this key point, which means that this first keyword is very important. Therefore, the first keywords corresponding to key points with a number of lines greater than the second threshold are identified as feature keywords. Feature keywords refer to important information in the contract. All feature keywords and their positions in the contract information structure diagram are recorded.

[0069] To improve the accuracy and readability of the contract information structure diagram, a standard thesaurus is pre-created, storing frequently queried standard words. For each keyword in the contract information structure diagram, the proximity between the standard words in the thesaurus is calculated and labeled. For each keyword, the first standard word with a proximity greater than a first threshold is selected. A proximity greater than the first threshold indicates that the first standard word and the keyword are relatively close and can be synonymously replaced. Then, the first standard word with the highest proximity to the keyword is selected as the second standard word, and the corresponding keyword is replaced with the second standard word. Figure 3 The diagram shown is the updated contract information structure. The second tag is the standard word that is closest to the keyword. Finally, the second standard word and the keyword are associated and saved to the first association table.

[0070] Since the standard terms are selected to be the closest to the keywords and also conform to search habits, the above method, by replacing the standard terms with keywords, can both preserve the original information of the contract to the greatest extent and improve the efficiency of contract content retrieval.

[0071] In one specific embodiment, finding and querying keywords that appear simultaneously based on historical query information includes the following steps:

[0072] All keywords in the historical query information are called fourth keywords. The fourth keywords are combined with the query keywords in pairs to generate multiple keyword pairs.

[0073] The historical query information is divided into multiple different sub-information sets according to time periods. The historical keyword groups in each sub-information set are extracted. For each keyword pair, the first occurrence in each historical keyword group is counted, as well as the second and third occurrences.

[0074] Compare the second and third counts, find the maximum value, add all the first counts together to get the first result value, divide the first result value by the maximum value to get the second result value, find the fourth keyword corresponding to the second result value that is greater than the second threshold, and use the fourth keyword as the same frequency keyword of the corresponding query keyword.

[0075] Specifically, to improve the comprehensiveness of the query, after obtaining the query keywords, keywords that frequently appear together with the current query keywords are extracted from historical query information. These keywords are called co-frequency keywords. First, all keywords in the historical query information are extracted and called fourth keywords. The fourth keywords and query keywords are then combined in pairs to generate multiple keyword pairs. Then, the historical query information is divided into multiple different sub-information sets according to time periods. For example, the historical query information entered within every half hour or ten minutes is a sub-information set. The historical keyword groups in each sub-information set are obtained. For the two keywords in each keyword pair, the first count of their simultaneous appearance in the historical keyword group is counted. The second count refers to the number of times the fourth keyword in the keyword pair appears in the historical keyword group. The third count refers to the number of times the query keyword in the keyword pair appears in the historical keyword group. Then, the second count and the third count are compared and the maximum value is obtained. A first count refers to the number of times the keyword pair appears simultaneously in a historical keyword group. All first counts are added together to obtain the total number of times the keyword pair appears simultaneously in all historical keyword groups. The first result value is divided by the maximum value to obtain the second result value. The fourth keyword in the keyword pair corresponding to the second result value that is greater than the second threshold is obtained, and the fourth keyword is used as the co-frequency keyword of the corresponding query keyword.

[0076] In one specific embodiment, extracting the first information and the second information from the contract information structure diagram includes the following steps:

[0077] Search and query the standard words that match the keywords from the first association table. If the standard words that match are found, obtain the candidate keywords based on the first association table.

[0078] Determine whether there are characteristic keywords that match the candidate keywords. If there are, obtain the position of the characteristic keywords in the contract information structure diagram and the corresponding key points. If not, find the key points that match the candidate keywords from the contract information structure diagram.

[0079] Based on the connecting lines of key points, related key points are found, and the found key points, related key points and corresponding connecting lines are used as the first information.

[0080] Specifically, to quickly retrieve contract content matching the query keywords, the system first searches the first association table for standard words that match the query keywords. Since standard words conform to user query habits, and the first association table stores all standard words related to contract content, the system first searches for corresponding standard words in the first association table. If a matching standard word is found, candidate keywords are obtained based on the first association table. Candidate keywords are keywords extracted from the contract text that correspond to the standard words. Since feature keywords are marked important keywords, the system first determines whether there are feature keywords that match the candidate keywords. If so, the feature keywords are retrieved from the contract information. The key points in the diagram are located and their corresponding key points. Related key points are then found based on the connecting lines of these key points. The found key points, related key points, and their corresponding connecting lines are marked as primary information. If no such key points are found, it indicates a mismatch between the query keywords and the feature keywords. In this case, key points matching the candidate keywords are directly searched from the contract information structure diagram. Then, related key points are found based on the connecting lines of these key points. The found key points, related key points, and their corresponding connecting lines are marked as primary information. Each key point and connecting line has corresponding keywords. These keywords are contract content that matches or is related to the query keywords; therefore, this information is used as primary information.

[0081] It is important to note that the search for standard words matching the query keywords in the first association table can be done by comparing the similarity between the standard words in the first association table and the query keywords. For example, calculate the similarity between the standard words in the first association table and the query keywords. If there are standard words in the first association table with a similarity greater than a first threshold, it means that there are standard words in the first association table that match the query keywords. Then, select the standard word with the highest similarity from the standard words that meet the conditions as the final standard word that matches the query keywords. If the similarity between all standard words and the query keywords is less than the first threshold, it means that there are no standard words that match the query keywords.

[0082] In one specific embodiment, when no matching standard word can be found, the following steps are included:

[0083] Search for standard words that match the same frequency keywords in the first association table. If matching standard words are found, obtain candidate keywords based on the first association table.

[0084] Determine whether there are characteristic keywords that match the same frequency keywords. If there are, obtain the position of the characteristic keywords in the contract information structure diagram and the corresponding key points. If not, search for key points that match the candidate keywords in the contract information structure diagram.

[0085] Based on the connecting lines of key points, related key points are found, and the found key points, related key points, and corresponding connecting lines are used as secondary information.

[0086] Specifically, if no standard word matching the query keyword is found, it means that there may be no content in the contract that matches the query keyword. However, in order to provide users with comprehensive query content, the same frequency keyword is a keyword that appears frequently at the same time as the query keyword. It is inferred that the user may search for content related to the same frequency keyword next. Therefore, based on the above steps, the second information matching the same frequency keyword is searched from the contract information structure diagram.

[0087] In one specific embodiment, displaying query results based on the first information and the second information includes the following steps:

[0088] Extract duplicate information from the first and second information, place the duplicate information in the first position of the query results, and sort and display other information based on its proximity to the query keywords.

[0089] Specifically, to enable users to intuitively obtain query results, duplicate information between the first and second pieces of information is extracted. Since the first piece of information is content directly matched by the query keywords, and the second piece of information is content that the user might query next, if there is duplicate information between the first and second pieces of information, it is inferred that the duplicate information is likely the content the user most wants to find. Therefore, the duplicate information is placed in the first position of the query results, that is, the first position, so that users can see it intuitively. Then, other information besides the duplicate information is sorted and displayed based on its proximity to the query keywords. Through the above steps, not only can users find content directly related to the query keywords, but also other related information can be found, helping users find more comprehensive and accurate results.

[0090] The above describes the AI-based contract content query method in the embodiments of this application. The following describes the AI-based contract content query system in the embodiments of this application. Please refer to [link / reference]. Figure 4 One embodiment of the contract content query system based on artificial intelligence in this application includes:

[0091] The structure diagram generation unit is used to split the contract text into multiple text blocks, perform semantic parsing on each text block to extract keywords, including the first keyword and the second keyword, and construct a contract information structure diagram based on the keywords;

[0092] The structure graph update unit is used to obtain the word vector of each keyword based on the pre-trained first model, calculate the proximity between every two keywords in the text block based on the word vector, improve the contract information structure graph based on the proximity, and create the first association table.

[0093] The homonym search unit is used to store historical query information, obtain query keywords based on real-time query statements, and search for homonyms that appear at the same time as the query keywords based on historical query information.

[0094] The result query unit is used to extract first information and second information from the contract information structure diagram based on query keywords, homonymous keywords and the first association table, and to display the query results based on the first information and second information.

[0095] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0096] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0097] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for querying contract content based on artificial intelligence, characterized in that, The method includes: The contract text is split into multiple text blocks, and semantic parsing is performed on each text block to extract keywords. The keywords include the first keyword and the second keyword. A contract information structure diagram is constructed based on the keywords. The word vectors of each keyword are obtained based on the pre-trained first model. The proximity between each pair of keywords in the text block is calculated based on the word vectors. The contract information structure graph is improved based on the proximity and a first association table is created. Store historical query information, obtain query keywords based on real-time query statements, and find keywords with the same frequency that appear at the same time as the query keywords based on historical query information; Based on query keywords, homogeneous keywords, and the first association table, first and second information are extracted from the contract information structure diagram, and query results are displayed based on the first and second information. Specifically, the method of improving the contract information structure diagram based on proximity includes: combining first keywords with a proximity greater than a preset first threshold to generate a first keyword group; selecting a keyword from the first keyword group as a target keyword; replacing the first keyword in the contract information structure diagram that belongs to the same first keyword group as the target keyword with the target keyword; combining second keywords with a proximity greater than the first threshold to generate a second keyword group; generating a third keyword based on the second keyword group; adding a corresponding key point in the contract information structure diagram based on the third keyword; and adding a second connecting line between the key point corresponding to the third keyword and the keywords in the second keyword group. Specifically, the method of improving the contract information structure diagram based on proximity includes: obtaining the number of first connecting lines connected to each key point; identifying the first keyword corresponding to the key point with a number of lines greater than a preset second threshold as a feature keyword; recording all feature keywords and their positions in the contract information structure diagram; pre-creating a standard thesaurus containing frequently queried standard words; calculating the proximity between each keyword in the contract information structure diagram and the standard words in the standard thesaurus; obtaining the first standard word with a proximity greater than a first threshold; obtaining the first standard word with the greatest proximity as the second standard word; replacing the corresponding keyword with the second standard word; and associating the second standard word and the keyword and saving them in a first association table.

2. The method according to claim 1, characterized in that, Construct a contract information structure diagram based on keywords, including: The first keyword is used as the key point in the contract information structure diagram. It is determined whether there is a relationship between any two first keywords. If so, the first connecting line is added to the two key points corresponding to the two first keywords that are related, and the relationship between the two first keywords is marked.

3. The method according to claim 1, characterized in that, Based on historical query information, search for keywords that appear simultaneously with the search keywords, including: All keywords in the historical query information are called fourth keywords. The fourth keywords are combined with the query keywords in pairs to generate multiple keyword pairs. The historical query information is divided into multiple different sub-information sets according to time periods. The historical keyword groups in each sub-information set are extracted. For each keyword pair, the first occurrence in each historical keyword group is counted, as well as the second and third occurrences. Compare the second and third counts, find the maximum value, add all the first counts together to get the first result value, divide the first result value by the maximum value to get the second result value, find the fourth keyword corresponding to the second result value that is greater than the second threshold, and use the fourth keyword as the same frequency keyword of the corresponding query keyword.

4. The method according to claim 1, characterized in that, Extract the first and second information from the contract information structure diagram, including: Search and query the standard words that match the keywords from the first association table. If the standard words that match are found, obtain the candidate keywords based on the first association table. Determine whether there are characteristic keywords that match the candidate keywords. If there are, obtain the position of the characteristic keywords in the contract information structure diagram and the corresponding key points. If not, find the key points that match the candidate keywords from the contract information structure diagram. Based on the connecting lines of key points, related key points are found, and the found key points, related key points and corresponding connecting lines are used as the first information.

5. The method according to claim 4, characterized in that, If no matching standard word is found, execute: Search for standard words that match the same frequency keywords in the first association table. If matching standard words are found, obtain candidate keywords based on the first association table. Determine whether there are characteristic keywords that match the same frequency keywords. If there are, obtain the position of the characteristic keywords in the contract information structure diagram and the corresponding key points. If not, search for key points that match the candidate keywords in the contract information structure diagram. Based on the connecting lines of key points, related key points are found, and the found key points, related key points, and corresponding connecting lines are used as secondary information.

6. The method according to claim 1, characterized in that, The query results are displayed based on the first and second information, including: Extract duplicate information from the first and second information, place the duplicate information in the first position of the query results, and sort and display other information based on its proximity to the query keywords.

7. An artificial intelligence-based contract content query system, used to implement the artificial intelligence-based contract content query method as described in any one of claims 1-6, characterized in that, The system includes: The structure diagram generation unit is used to split the contract text into multiple text blocks, perform semantic parsing on each text block to extract keywords, including the first keyword and the second keyword, and construct a contract information structure diagram based on the keywords; The structure graph update unit is used to obtain the word vector of each keyword based on the pre-trained first model, calculate the proximity between every two keywords in the text block based on the word vector, improve the contract information structure graph based on the proximity, and create the first association table. The homonym search unit is used to store historical query information, obtain query keywords based on real-time query statements, and search for homonyms that appear at the same time as the query keywords based on historical query information. The result query unit is used to extract first information and second information from the contract information structure diagram based on query keywords, homonymous keywords and the first association table, and to display the query results based on the first information and second information.

Images