Legal document processing method and system based on big data

Abstract

This application relates to the field of text processing technology, and in particular to a method and system for processing legal documents based on big data. The method includes: segmenting a case description to obtain multiple descriptive words; calculating the matching effectiveness of each descriptive word in a case library; using the product of the normalized matching effectiveness and the TF-IDF value of the descriptive word as a weighting coefficient to weight the semantic vectors of each descriptive word to obtain case characteristics; and obtaining similar cases to the case description based on the similarity between the case characteristics and the case characteristics of historical cases. The technical solution of this application can improve the accuracy of similar case retrieval results.

Description

Technical Field

[0001] The present application relates to the field of text processing technology, and in particular to a legal document processing method and system based on big data. Background Art

[0002] As legal education continues to gain wider recognition and understanding, more and more people are turning to online searches for legal solutions when faced with problems. This has led to a growing demand for legal case searches. As the number of cases increases, the challenge of accurately capturing the text features of the legal documents corresponding to each case, and thus enabling accurate retrieval of similar cases, is a pressing issue.

[0003] At present, the patent application document with publication number CN110928994A discloses a similar case retrieval method, a similar case retrieval device and an electronic device, wherein the method includes: receiving a case to be retrieved, the case to be retrieved includes at least one of a text description and a multimedia file related to the case; performing content analysis on the text description to perform paragraph identification and performing dispute focus analysis, legal element analysis, keyword extraction, multi-model semantic processing and multi-granularity semantic processing on the identified paragraph to generate a document analysis result of the case to be retrieved, wherein the identified paragraph is subjected to multi-model semantic processing; performing semantic processing on the multimedia file to generate a semantic analysis result of the case to be retrieved; performing multi-model semantic, multi-granularity and multi-modal semantic matching based on the document analysis result and semantic analysis result of the case to be retrieved and the document analysis result and semantic analysis result of the case in the case library to obtain a retrieval result.

[0004] The above method realizes similar case retrieval by extracting the document parsing results and semantic parsing results of the case to be retrieved. When processing the text description, multiple models are used to extract features from the paragraphs in the text description. However, document parsing is performed on a paragraph basis, ignoring the impact of keywords such as "civil" and "criminal" in legal documents on case similarity, resulting in inaccurate retrieval results for similar cases. Summary of the Invention

[0005] In order to solve the technical problem of inaccurate retrieval results of similar cases, the present application provides a legal document processing method and system based on big data, which can improve the accuracy of similar case retrieval results.

[0006] In a first aspect, the present application provides a legal document processing method based on big data, the search method comprising: performing word segmentation on a case description to obtain a plurality of descriptive words; calculating the matching validity of each descriptive word in a case library, comprising: the case library comprising a plurality of case pairs with matching labels, the matching labels comprising matching and non-matching; counting a first co-occurrence probability of any descriptive word in case pairs whose matching labels are matching; counting a second co-occurrence probability of the descriptive word in case pairs whose matching labels are non-matching, and taking a ratio of the first co-occurrence probability to the second co-occurrence probability as the matching validity of the descriptive word; taking the product of the normalized matching validity and the TF-IDF value of the descriptive word as a weighting coefficient to weightedly sum the semantic vectors of each descriptive word to obtain case features, and obtaining similar cases of the case description based on the similarity between the case features and the case features of historical cases.

[0007] The case description is segmented to obtain multiple descriptive words for the case description; the case library includes multiple case pairs, which can be divided into matching case pairs and non-matching case pairs; the first co-occurrence probability of any descriptive word is counted in the matching case pairs; the second co-occurrence probability of the descriptive word is counted in the non-matching case pairs, and the ratio of the first co-occurrence probability to the second co-occurrence probability is used as the matching validity of the descriptive word, which can measure the ability of the descriptive word to judge whether the legal documents are similar; further, the product of the normalized matching validity and the TF-IDF value is used as the weighting coefficient to weight the sum of the semantic vectors of each descriptive word to obtain the case feature, and the weighting coefficient of each descriptive word is determined by comprehensively considering the matching validity and TF-IDF value, so as to accurately extract the case feature while ensuring that the extracted case feature can accurately judge whether the case feature is similar to the historical case; finally, based on the similarity between the case feature and the case feature of the historical case, similar cases of the case description are obtained to ensure the accuracy of the similar case retrieval results.

[0008] Preferably, the case description is segmented using jieba segmentation.

[0009] Preferably, the descriptor Matching effectiveness for:

[0010] , Descriptive words The first co-occurrence probability of Descriptive words The second co-occurrence probability of .

[0011] Preferably, the method for obtaining the semantic vector of the description word includes: in response to the matching validity of the description word being greater than a preset threshold, using the word vector of the description word as the semantic vector; otherwise, using the sum of the word vectors of the description word and multiple description words in the context information as the semantic vector of the description word.

[0012] For descriptive words that play a positive role in determining whether case pairs are similar (i.e., descriptive words with a matching validity greater than 1), their own word vectors can already distinguish whether case pairs are similar, and the word vectors can be directly used as semantic vectors. However, for descriptive words that do not play a role or play a negative role in determining whether case pairs are similar (i.e., descriptive words with a matching validity less than or equal to 1), their own word vectors cannot distinguish whether case pairs are similar. In this case, it is necessary to combine the contextual information of the case description to accurately obtain the semantic vector of the descriptive word.

[0013] Preferably, the word vector is obtained using the Legal-BERT or Word2Vec model.

[0014] Preferably, taking the sum of the word vectors of the descriptive word and multiple descriptive words in the context information as the semantic vector of the descriptive word includes: setting an initial window, the initial window includes the context information of the descriptive word; in the case pair that also contains the descriptive word, taking the sum of the word vectors in the initial window as the semantic vector of the descriptive word, and calculating the minimum Euclidean distance of the semantic vectors of the descriptive word between different historical cases in the case pair as the semantic deviation of the descriptive word; adjusting the size of the initial window, and taking the initial window corresponding to the maximum value of the objective function as the target window, the objective function is positively correlated with the semantic deviation of the descriptive word in the unmatched case pair, and negatively correlated with the semantic deviation of the descriptive word in the matched case pair; extracting the semantic vector of the descriptive word in the case description based on the target window.

[0015] The initial window includes the contextual information of the descriptive word. By adjusting the size of the initial window, more contextual information of the descriptive word is continuously introduced until the semantic information of the descriptive word can distinguish whether the case pairs are similar. Then, the target window is obtained, and the accurate extraction of the semantic information of the descriptive word in the legal document is achieved. The extracted semantic information can effectively distinguish whether the case pairs are similar.

[0016] Preferably, the objective function Satisfies the relationship: ; For unmatched case pairs The semantic deviation of the descriptive words described in For matching case pairs The semantic deviation of the descriptive words described in and They are the set of unmatched case pairs and the set of matched case pairs, respectively.

[0017] Objective function The value of can accurately reflect the ability of the semantic vector of the descriptive word in the initial window to judge whether the case pairs are similar.

[0018] Preferably, after obtaining similar cases of the case description, the processing method also includes: determining the scores of similar cases based on user feedback information, and in response to the score of any similar case being greater than a score threshold, treating the case description and the similar case as a group of case pairs with matching labels; otherwise, treating the case description and the similar case as a group of case pairs with mismatching labels to update the case library.

[0019] Realize continuous updating of the case library and gradually optimize the matching validity calculation to ensure the accuracy of similar cases.

[0020] Preferably, user feedback information includes adoption, collection and likes.

[0021] In a second aspect of the present application, a legal document processing system based on big data is also provided, comprising a processor and a memory, wherein the memory stores computer program instructions. When the computer program instructions are executed by the processor, a legal document processing method based on big data as described in the first aspect of the present application is implemented.

[0022] The technical solution of this application has the following beneficial technical effects:

[0023] The case description is segmented to obtain multiple descriptive words for the case description; the case library includes multiple case pairs, and the case pairs can be divided into matching case pairs and non-matching case pairs; the first co-occurrence probability of any descriptive word is counted in the matching case pairs; the second co-occurrence probability of the descriptive word is counted in the non-matching case pairs, and the ratio of the first co-occurrence probability to the second co-occurrence probability is used as the matching validity of the descriptive word. The matching validity can measure the ability of the descriptive word to judge whether the legal documents are similar. When the descriptive word plays a positive role in judging whether the case pairs are similar, the matching validity of the descriptive word is greater than 1. When the descriptive word does not play a role in judging whether the case pairs are similar, the matching validity of the descriptive word is greater than 1. The matching validity is equal to 1. When the descriptive word plays a reverse role in judging whether the case pair is similar, the matching validity of the descriptive word is less than 1. Furthermore, the product of the normalized matching validity and the TF-IDF value is used as the weighting coefficient to weight the sum of the semantic vectors of each descriptive word to obtain the case features. The weighting coefficient of each descriptive word is determined by comprehensively considering the matching validity and TF-IDF value. While accurately extracting the case features, it is ensured that the extracted case features can accurately judge whether the case features are similar to historical cases. Finally, based on the similarity between the case features and the case features of historical cases, similar cases described in the case are obtained to ensure the accuracy of the similar case retrieval results. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is a flowchart of a legal document processing method based on big data according to an embodiment of the present application.

[0025] Figure 2 It is a block diagram of a legal document processing system based on big data according to an embodiment of the present application. Specific implementation manners

[0026] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0027] According to the first aspect of the present application, the present application provides a legal document processing method based on big data for retrieving similar cases according to case descriptions. Figure 1 It is a flowchart of a legal document processing method based on big data according to an embodiment of the present application. As Figure 1 shown, the legal document processing method based on big data includes steps S101 to S103, which are described in detail below.

[0028] S101, perform word segmentation on the case description to obtain multiple description words.

[0029] In one embodiment, obtain the case description input by the user, perform word segmentation on the case description using jieba word segmentation, and remove stop words after the word segmentation to obtain multiple description words; wherein, the stop words are function words or modal particles such as "de", "qi", "a", etc.

[0030] Among them, jieba word segmentation is an existing Chinese word segmentation tool that can split Chinese text into single words, and one word is a description word.

[0031] S102, calculate the matching effectiveness of each description word in the case library.

[0032] In one embodiment, the matching effectiveness of a description word can measure the ability of the description word to judge whether legal documents are similar. The greater the matching effectiveness, the stronger the ability of the description word to judge whether legal documents are similar.

[0033] Specifically, calculating the matching effectiveness of each description word in the case library includes: the case library includes multiple case pairs with matching labels, and the matching labels include matching and non - matching; count the first co - occurrence probability of any description word in the case pairs with the matching label being matching; count the second co - occurrence probability of the description word in the case pairs with the matching label being non - matching, and take the ratio of the first co - occurrence probability to the second co - occurrence probability as the matching effectiveness of the description word.

[0034] Among them, the case library includes multiple case pairs, one case pair includes two historical cases, and one historical case corresponds to one or more legal documents; one case pair corresponds to a matching label, and the matching label is manually marked, including matching and mismatching. If the matching label is matching, the similarity between the two historical case legal documents in the case pair is considered to be 1; if the matching label is mismatching, the similarity between the two historical case legal documents in the case pair is considered to be 0.

[0035] For any descriptor, there are three cases:

[0036] The first case: The first co-occurrence probability is large and the second co-occurrence probability is small, indicating that the probability of the descriptive word appearing in the matching case pair is greater than the probability of appearing in the non-matching case pair. The descriptive word can effectively distinguish the matching case pairs from the non-matching case pairs, and plays a positive role in determining whether the case pairs are similar. The matching effectiveness of the descriptive word should be a large value.

[0037] The second case: The first co-occurrence probability is equal to the second co-occurrence probability, indicating that the descriptive word appears the same number of times in the matching case pairs and the non-matching case pairs. The descriptive word cannot effectively distinguish the matching case pairs from the non-matching case pairs and does not play a role in determining whether the case pairs are similar.

[0038] The third case: the first co-occurrence probability is small and the second co-occurrence probability is large, which means that the probability of the descriptive word appearing in the unmatched case pairs is greater than the probability of appearing in the matched case pairs. In other words, in the unmatched case pairs, the descriptive word also has a large co-occurrence probability. Although the descriptive word can distinguish between matched case pairs and unmatched case pairs, it will mistakenly judge the unmatched case pairs as matched case pairs. Therefore, the descriptive word plays a reverse role in the process of judging whether the case pairs are similar, and the matching validity of the descriptive word should be a small value.

[0039] Specifically, the descriptive word Matching effectiveness Satisfies the following formula:

[0040] , Descriptive words The first co-occurrence probability of Descriptive words The second co-occurrence probability of .

[0041] In summary, for any descriptive word obtained in step S101, the matching validity of the descriptive word can be obtained. When the descriptive word plays a positive role in judging whether the case pairs are similar, the matching validity of the descriptive word is greater than 1. When the descriptive word does not play a role in judging whether the case pairs are similar, the matching validity of the descriptive word is equal to 1. When the descriptive word plays a negative role in judging whether the case pairs are similar, the matching validity of the descriptive word is less than 1, thereby achieving accurate quantification of the matching validity of each descriptive word.

[0042] S103: The product of the normalized matching validity and the TF-IDF value is used as a weighting coefficient to weight the sum of the semantic vectors of each descriptive word to obtain case features. Based on the similarity between the case features and the case features of historical cases, similar cases of case descriptions are obtained.

[0043] In one embodiment, the word vector of each description word is directly used as the semantic vector of the corresponding description word.

[0044] It should be noted that for descriptive words that play a positive role in determining whether case pairs are similar (i.e., descriptive words with a matching validity greater than 1), their own word vectors can already distinguish whether case pairs are similar. However, for descriptive words that do not play a role or play a negative role in determining whether case pairs are similar (i.e., descriptive words with a matching validity less than or equal to 1), their own word vectors cannot distinguish whether case pairs are similar. In this case, it is necessary to combine the contextual information of the case description to determine the semantic vector of the descriptive word.

[0045] Specifically, the method for obtaining the semantic vector of the description word includes: in response to the matching validity of the description word being greater than a preset threshold, using the word vector of the description word as the semantic vector; otherwise, using the sum of the word vectors of the description word and multiple description words in the context information as the semantic vector of the description word.

[0046] The preset threshold value is 1.

[0047] In one embodiment, the sum of the word vectors of the description word and multiple description words in the context information is used as the semantic vector of the description word, including: obtaining the word vector of the description word; setting an initial window, the initial window including the context information of the description word; in a case pair that also contains the description word, the sum of the word vectors in the initial window is used as the semantic vector of the description word, and the minimum Euclidean distance of the semantic vectors of the description word between different historical cases in the case pair is calculated as the semantic deviation of the description word; adjusting the size of the initial window, and using the initial window corresponding to the maximum value of the objective function as the target window, the objective function is positively correlated with the semantic deviation of the unmatched case pair and negatively correlated with the semantic deviation of the matched case pair; extracting the semantic vector of the description word in the case description based on the target window.

[0048] Among them, the word vector can be obtained using the Legal-BERT or Word2Vec model. Legal-BERT is a pre-trained model in the legal field, which is used to obtain the semantic vector of each descriptive word in the legal document. It is a well-known technology for those skilled in the art and will not be described here.

[0049] Among them, the size of the initial window is 3, including one word each of the description word and the description word context; when adjusting the size of the initial window, the initial window is expanded by a distance of one word toward the context each time, that is, after completing one size adjustment, the size of the initial window is 5, including two words each of the description word and the description word context.

[0050] A case pair includes two historical cases, which are recorded as Historical Case 1 and Historical Case 2 respectively. If the legal documents of the two historical cases both include the descriptive words, the case pair is regarded as a case pair that simultaneously contains the descriptive words. The semantic vectors of the descriptive words in the legal documents of Historical Case 1 and Historical Case 2 are obtained respectively. Since the descriptive words appear at least once in Historical Case 1 and Historical Case 2, the minimum Euclidean distance between the semantic vectors of Historical Case 1 and Historical Case 2 is taken as the semantic deviation in the case pair.

[0051] By adjusting the size of the initial window, more contextual information about the descriptive word is continuously introduced, and the semantic information of the descriptive word in the legal document is obtained more accurately. After each adjustment, an initial window and the objective function value of the initial window are obtained. The objective function is positively correlated with the semantic deviation of the unmatched case pairs and negatively correlated with the semantic deviation of the matched case pairs, which can reflect the ability of the semantic vector of the descriptive word to judge whether the case pairs are similar. Specifically, the objective function Satisfies the relationship:

[0052] ; For unmatched case pairs The semantic deviation of the descriptive words described in For matching case pairs The semantic deviation of the descriptive words described in and They are the set of unmatched case pairs and the set of matched case pairs, respectively.

[0053] As can be understood, when the objective function reaches its maximum value, it indicates that the initial window at that point maximizes the ability of the descriptor to determine case similarity. This initial window is then used as the target window for that descriptor, and is used to extract the semantic vector for that descriptor in the case description. An optimization algorithm, such as a simulated annealing algorithm or a hill climbing algorithm, can be used to determine the target window.

[0054] In this way, the semantic vector of each descriptive word in the case description is obtained, and the product of the normalized matching validity and the TF-IDF value is used as the weighting coefficient. The semantic vectors of each descriptive word are weighted and summed according to the weighting coefficient to obtain the case characteristics. The case characteristics focus on the descriptive words that play a positive role in determining whether the case pairs are similar, and weaken the descriptive words that play a negative role.

[0055] The ratio of the matching validity of any descriptive word to the sum of the matching validity of all descriptive words in the case description is used as the normalized matching validity of the descriptive word.

[0056] Among them, the TF-IDF algorithm is a commonly used technical means in text processing. It obtains the TF-IDF value of each descriptive word in the legal documents of all historical cases. The TF-IDF value tends to filter out common descriptive words and retain important descriptive words; the weighting coefficient of each descriptive word is determined by comprehensively considering the matching effectiveness and TF-IDF value, accurately extracting case features while ensuring that the extracted case features can accurately determine whether the case features are similar to historical cases.

[0057] In one embodiment, case features of each historical case are obtained using the same method, and the similarity between the case features and the case features of the historical case is calculated. The cases are then ranked in descending order of similarity, and the top-ranked historical cases are designated as similar cases to the case description. The similarity is calculated using a Euclidean distance-based similarity method. If the Euclidean distance between the case features and the case features of any historical case is large, the case description is less similar to the historical case.

[0058] In one embodiment, after obtaining similar cases of the case description, the processing method further includes: determining the scores of similar cases based on user feedback information, and in response to any similar case having a score greater than a score threshold, treating the case description and the similar case as a group of case pairs with a matching label; otherwise, treating the case description and the similar case as a group of case pairs with a mismatching label, thereby updating the case library.

[0059] User feedback can include user behaviors such as adoption, favorites, and likes. Each behavior is assigned a score. For example, adoption is scored as 4, favorites as 3, and likes as 3, for a total score of 10. When a user performs a corresponding behavior, they receive the corresponding score. The score threshold is 5. When the score of any similar case exceeds the threshold, the case description and the similar case are considered a pair with a matching label. When the score of any similar case does not exceed the threshold, the case description and the similar case are considered a pair with a mismatching label. The case library is continuously updated, and the matching validity calculation is gradually optimized to ensure the accuracy of similar cases.

[0060] According to the second aspect of the present application, the present application also provides a legal document processing system based on big data. Figure 2 This is a structural diagram of a legal document processing system based on big data according to an embodiment of the present application. Figure 2 As shown, the system 50 includes a processor and a memory. The memory stores computer program instructions. When executed by the processor, the computer program instructions implement the big data-based legal document processing method according to the first aspect of this application. The system also includes other components familiar to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are well known in the art and are therefore not described in detail here.

[0061] It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present application, and these all fall within the scope of protection of the present application.

Claims

1. A legal document processing method based on big data, characterized in that: The processing method comprises: Segment the case description to obtain multiple descriptive words; Calculating the matching validity of each descriptive word in the case library, including: the case library includes a plurality of case pairs with matching labels, wherein the matching labels include matching and non-matching; counting the first co-occurrence probability of any descriptive word in the case pairs with matching labels; counting the second co-occurrence probability of the descriptive word in the case pairs with non-matching labels, and taking the ratio of the first co-occurrence probability to the second co-occurrence probability as the matching validity of the descriptive word; the descriptive word Matching effectiveness for: , Descriptive words The first co-occurrence probability of Descriptive words The second co-occurrence probability of The product of the normalized matching validity and the TF-IDF value of the descriptive word is used as a weighting coefficient to weight the sum of the semantic vectors of each descriptive word to obtain the case characteristics. Based on the similarity between the case characteristics and the case characteristics of historical cases, similar cases of case descriptions are obtained.

2. The method for processing legal documents based on big data according to claim 1, characterized in that: Use jieba segmentation to segment the case description.

3. The method for processing legal documents based on big data according to claim 1, characterized in that: The method for obtaining the semantic vector of the description word includes: In response to the matching validity of the description word being greater than a preset threshold, the word vector of the description word is used as a semantic vector. Otherwise, the sum of the word vectors of the description word and multiple description words in the context information is used as the semantic vector of the description word.

4. The method for processing legal documents based on big data according to claim 3, characterized in that: The word vector is obtained using the Legal-BERT or Word2Vec model.

5. The method for processing legal documents based on big data according to claim 3, characterized in that: Taking the sum of the word vectors of the description word and multiple description words in the context information as the semantic vector of the description word includes: Setting an initial window, wherein the initial window includes context information of the description word; In case pairs that both contain the descriptive word, the sum of the word vectors in the initial window is used as the semantic vector of the descriptive word, and the minimum Euclidean distance between the semantic vectors of the descriptive word between different historical cases in the case pair is calculated as the semantic deviation of the descriptive word; The size of the initial window is adjusted, and the initial window corresponding to the maximum value of the objective function is used as the target window. The objective function is positively correlated with the semantic deviation of the descriptive words in the unmatched case pairs and negatively correlated with the semantic deviation of the descriptive words in the matched case pairs. The semantic vectors of the descriptive words in the case description are extracted based on the target window.

6. The method for processing legal documents based on big data according to claim 5, characterized in that: The objective function Satisfies the relationship: ; For unmatched case pairs The semantic deviation of the descriptive words described in For matching case pairs The semantic deviation of the descriptive words described in and They are the set of unmatched case pairs and the set of matched case pairs, respectively.

7. The method for processing legal documents based on big data according to claim 1, characterized in that: After obtaining a similar case as described in the case description, the processing method further includes: The scores of similar cases are determined based on user feedback information. In response to the score of any similar case being greater than a score threshold, the case description and the similar case are regarded as a group of case pairs with a matching label. Otherwise, the case description and the similar case are regarded as a group of case pairs with a matching label, thereby updating the case library.

8. The method for processing legal documents based on big data according to claim 7, characterized in that: User feedback information includes adoption, collection and likes.

9. Legal document processing system based on big data, characterized by: The method comprises a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the method for processing legal documents based on big data according to any one of claims 1 to 8 is implemented.

Images