Legal provision retrieval method and system

Abstract

The invention relates to the field of electronic digital data processing, in particular to a legal provision retrieval method and system.The method comprises the steps that a related entry set in a Boolean model is obtained, and a preliminary retrieval result and a final retrieval result are obtained through retrieval in a database; according to the classification information of the vocabulary entries, counting the occurrence frequency of each vocabulary entry in different classifications, calculating the correlation probability in each classification, and selecting the maximum correlation probability as the importance weight of the vocabulary entry; calculating the correlation degree between any two entries in the related entry set, weakening the entry combination with high occurrence frequency, and accumulating the weakening degrees to obtain the total weakening degree; and performing weighted summation on the importance weight and the occurrence frequency of the entries in the legal provisions, and performing correction according to the total weakening degree to obtain the priority degree of a final retrieval result. According to the method, the relevance and the accuracy of the retrieval results are improved by weakening the retrieval results and sorting the priority degrees of the retrieval results.

Description

Technical Field

[0001] The present invention relates to the field of electronic digital data processing, and in particular to a legal text retrieval method and system. Background Art

[0002] Legal text retrieval refers to the process of quickly and accurately finding legal texts or information related to user needs in massive legal documents (such as laws and regulations, judicial interpretations, cases, academic articles, etc.) through specific technologies or methods.

[0003] With the increasing complexity of the legal system and the continuous updating of legal provisions, when encountering various problems in life and needing to maintain one's rights and interests through legal provisions, people usually search for relevant legal provisions on the Internet. The Internet has greatly improved the accessibility of legal information. People can quickly find relevant legal provisions, judicial interpretations and cases through search engines, legal databases, official government websites and other channels. Therefore, the Internet has made legal knowledge no longer limited to legal practitioners. Ordinary citizens can also understand their rights and obligations through online searches, thereby better safeguarding their legitimate rights and interests.

[0004] At present, legal issues often involve legal provisions in multiple fields. In addition to national laws, there are also multi-level legal norms such as local regulations and judicial interpretations. These legal norms complement each other and also increase the complexity of retrieval. With the complexity of the legal system and the continuous updating of legal provisions, traditional Boolean model retrieval methods face many challenges when dealing with large-scale, multi-field and multi-level legal norms. In the process of searching legal provisions using conventional Boolean models, the search results of each term are directly output, resulting in users still needing to find the text they want from a large number of search results after searching, which reduces the accuracy and convenience of Boolean model queries. Summary of the invention

[0005] In order to solve the problems of low retrieval efficiency, insufficient result accuracy and the need for users to manually screen a large amount of irrelevant information when traditional Boolean models are used to process large-scale, multi-field and multi-level legal norms, the present invention provides solutions in the following aspects.

[0006] In a first aspect, a method for searching legal texts includes: obtaining a set of related terms in a Boolean model, performing a database search on the terms in the set of related terms, obtaining a preliminary search result containing any term in the set of related terms, and screening out a search result that strictly matches the Boolean expression query condition from the preliminary search result as a final search result; obtaining classification information of the terms, counting the frequency of occurrence of each term in each classification information, calculating the relevance probability of the term in each classification information, and selecting the largest relevance probability in each classification information as the importance weight of the term, wherein the classification information refers to various classification attributes related to the legal text; calculating the degree of relevance between any two terms in the set of related terms, obtaining a combination of terms with a higher frequency of occurrence in the set of related terms, performing a weakening process based on the degree of relevance, obtaining the degree of weakening of each combination of terms and summing them up to obtain the overall degree of weakening of the preliminary search result; performing a weighted summation of the importance weight and the frequency of occurrence of the terms in the legal text, and performing correction based on the overall degree of weakening to obtain the priority of the final search result, sorting the final search result according to the priority, and obtaining the final display result.

[0007] By comprehensively considering the classification information, frequency of occurrence, related probability and the degree of relevance between terms, the matching degree between legal provisions and user query intentions is accurately evaluated, thereby significantly improving the relevance and accuracy of search results. At the same time, the search results are weighted and corrected based on the importance weight, frequency of occurrence and overall weakening degree of the terms, which optimizes the sorting of search results and avoids the false high priority caused by repeated weighting of terms, so as to make the result sorting more reasonable. Through precise screening and sorting, the time and energy of users in screening the required information from a large number of results is reduced, significantly improving user experience and search efficiency. The final display results are presented in a list sorted by priority, allowing users to quickly locate the most relevant legal provisions, meeting the user's needs for efficiency and convenience in legal retrieval.

[0008] Preferably, the preliminary search results are all legal provisions containing relevant terms retrieved based on the Boolean model, and the final search results are legal provisions that fully meet the user's search conditions screened from the preliminary search results; In response to the legal text containing any term in the Boolean model, the legal text is considered to be a preliminary search result. In response to the legal text containing all terms in the Boolean model, the legal text is considered to be a final detection result.

[0009] By clearly distinguishing between preliminary search results and final search results, accurate positioning and efficient screening of search results are achieved. Preliminary search results are based on the existence of any term in the Boolean model to ensure that the search scope is comprehensive and no potentially relevant clauses are missed; while the final search results strictly screen out clauses that meet all terms in the Boolean model, accurately match user search conditions, and improve the pertinence and accuracy of search results.

[0010] Preferably, the related probability includes: Taking any term as the target term, calculate the ratio between the sum of the frequency of occurrence of the target term in all legal articles in a specific classification and the sum of the frequency of occurrence of all legal articles in all classifications, and obtain the relative importance of the target term in the specific classification; Calculate the ratio of the number of legal articles of the target term in the specific classification information to the total number of search results to obtain the concentration of the target term in the specific classification information; The product of the relative importance and the concentration level is used as the relevance probability of the target term in the specific classification information.

[0011] By calculating the relative importance and concentration of the target term in a specific category and taking the product as the relevant probability, the importance and relevance of the term in different categories can be accurately evaluated. The relative importance reflects the activity of the term in the category, and the concentration reflects the concentrated distribution of the term in the category. It can effectively distinguish the actual contribution of the term in different categories, thereby improving the relevance and accuracy of the search results.

[0012] Preferably, the related probability further includes: Taking any term as the target term, the ratio between the sum of the occurrence frequencies of the target term in all legal provisions in the specific classification information and the total number of search results of the target term is calculated, and the ratio is taken as the relevant probability of the target term.

[0013] Preferably, the degree of correlation includes: Divide the number of times any two terms appear simultaneously in the preliminary search results by the total number of times the two terms appear in the preliminary search results respectively, and take the square root of the result. The square root is used as the correlation degree between the two terms.

[0014] By quantifying the strength of association between terms, we can more accurately evaluate the relevance between terms, thereby reasonably adjusting the weight of terms in the search results, avoiding inflated priorities due to repeated weighting of terms, improving the relevance and accuracy of search results, and ensuring that users can find the most relevant legal provisions more efficiently.

[0015] Preferably, the degree of weakening includes: Calculate the joint occurrence frequency of any two terms in the preliminary search results. If the joint frequency is greater than the average occurrence frequency of all terms, the relevance of the two terms is weakened; otherwise, if the joint frequency is less than or equal to the average occurrence frequency of all terms, no weakening is required. The weakening method is: using a negative exponential function to weaken the relevance of two terms, so as to reduce the priority of the search results.

[0016] By calculating the joint frequency of any two terms in the preliminary search results and comparing them with the average frequency of all terms, the relevance of the terms is dynamically adjusted. When the joint frequency is higher than the average, the negative exponential function is used to weaken the relevance of the two terms, thereby reducing the priority of the search results; conversely, if the joint frequency is lower than or equal to the average, the relevance remains unchanged. This dynamic adjustment mechanism can effectively avoid the false high priority caused by repeated weighting of terms, and ensure the accuracy and relevance of the search results.

[0017] Preferably, the overall weakening degree satisfies the following relationship: ; In the formula, Indicates The overall weakening of the initial search results, Represents the total number of terms in the related term set, Indicates The first search result Entries and The degree of weakening of an entry.

[0018] Preferably, the priority satisfies the following relationship: ; In the formula, Indicates The priority of the initial search results, Indicates The overall weakening of the initial search results, Represents the total number of terms in the related term set, Indicates The importance weight of each term, Indicates The final search results are selected from the preliminary search results. The frequency of occurrence of a term.

[0019] Preferably, obtaining the final display result includes: The final search results are sorted from largest to smallest based on priority and output in a list format for viewing the most relevant legal provisions.

[0020] In a second aspect, a legal text retrieval system includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned legal text retrieval method is implemented.

[0021] The present invention has the following effects: 1. The present invention calculates the relevance probability of entries in each category and the degree of relevance between entries, weakens the preliminary search results and sorts the final search results according to priority, so as to effectively screen out the legal provisions that best match the user's query intention, improve the relevance and accuracy of the search results, and reduce the time and energy of users in screening the required information from a large number of results.

[0022] 2. The present invention optimizes the search algorithm to reduce the falsely high priority caused by repeated weighting of terms, thereby ensuring the quality and reliability of the search results. At the same time, the final search results are sorted by priority and output in a list form, allowing users to quickly view the most relevant legal provisions, significantly improving user experience and search efficiency, and meeting the diverse needs of users in legal searches.

[0023] 3. The present invention can dynamically adjust the priority of search results by comprehensively considering multiple factors such as the classification information, frequency of occurrence, and relevance of the entries, thereby enhancing the adaptability of the search system to complex legal systems and multi-field legal norms, enabling it to better cope with the continuous updating and changes of legal provisions, and providing users with smarter and more accurate search services. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a method flow chart of steps S1 to S4 in a legal text retrieval method according to an embodiment of the present invention.

[0025] Figure 2 It is a structural block diagram of a legal text retrieval system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments.

[0027] Reference Figure 1 A legal text retrieval method includes steps S1 to S4, which are as follows: S1: Obtain a set of related terms in the Boolean model, perform a database search on the terms in the set of related terms, obtain a preliminary search result containing any term in the set of related terms, and filter out the search results that strictly match the Boolean expression query conditions from the preliminary search results as the final search results.

[0028] It should be noted that users enter keywords or phrases in the search box, and these input terms constitute a term set. The system automatically expands related synonyms, related words, professional terms, etc. based on the terms entered by the user to form a more comprehensive set of related terms.

[0029] Exemplarily, the user enters a keyword or phrase in the search box, for example, the user enters: "contract breach"; the system expands: the term set includes "contract breach", "breach of contract liability", "contract termination", "liquidated damages", etc.

[0030] The preliminary search results are all the legal provisions containing relevant terms retrieved based on the Boolean model, and the final search results are the legal provisions that fully meet the user's search conditions screened from the preliminary search results; In response to the legal text containing any term in the Boolean model, the legal text is considered to be a preliminary search result. In response to the legal text containing all terms in the Boolean model, the legal text is considered to be a final detection result.

[0031] For example, in the Boolean model, input: 1and2. Article 1: contains the term "1" but does not contain the term "2", which belongs to the preliminary search results but not the final search results; Article 2: contains both the term "1" and the term "2", which belongs to the preliminary search results but also belongs to the final search results.

[0032] It should be noted that the difference between the preliminary search results and the final search results is: preliminary search results: contain legal provisions of any term in the Boolean model, which is used to expand the search scope; final search results: contain legal provisions of all terms in the Boolean model, which is used to ensure the accuracy and relevance of the search results.

[0033] To further explain, each legal provision has its own classification information. Specifically, legal provisions are subdivided based on the legal field. For example, the classification information includes: civil law, criminal law, commercial law, administrative law and intellectual property law, etc. The total number of classifications of legal provisions is recorded as , get all search results for each term .

[0034] S2: Obtain the classification information of the term, count the frequency of occurrence of each term in each classification information, calculate the relevant probability of the term in each classification information, and select the largest relevant probability in each classification information as the importance weight of the term, where the classification information refers to various classification attributes related to legal provisions.

[0035] Related probabilities, including: Taking any term as the target term, calculate the ratio between the sum of the frequency of occurrence of the target term in all legal articles in a specific classification and the combined frequency of occurrence of the target term in all legal articles in all classifications, and obtain the relative importance of the target term in the specific classification; Calculate the ratio of the number of legal articles of the target term in the specific classification information to the total number of search results to obtain the concentration of the target term in the specific classification information; The product between the relative importance and the concentration is taken as the relevant probability of the target term in the specific classification information.

[0036] Specifically, the relevant probability satisfies the following relationship: ; In the formula, Indicates The entry is in The relevant probability of all search results of the class, Indicates The entry is in The number of class search results, Indicates The entry is in No. The frequency of occurrence of legal provisions, Indicates the total number of categories of legal provisions. Indicates The total number of search results for terms.

[0037] That is to say, Indicates The entry is in The total frequency of occurrence in the classification search results reflects the No. The total frequency of the entries is analyzed by analyzing the proportion of the frequency of each category to the frequency of all classified information. The higher the proportion, the higher the proportion. Entries and The greater the relevance of the category, the more Entries to distinguish Classification of legal provisions.

[0038] The relevance of a term in a classification is evaluated by the proportion of global occurrence frequency. If a term appears more frequently in a classification and less frequently in other classifications, the probability of relevance is higher. It reflects the concentration of the search results of the term in the classification information. If it is mainly concentrated in a certain classification information, the corresponding correlation probability will be higher. It is more suitable for evaluating the global relevance of the term in a certain classification information, and can better reflect the classification differences of the term in different classification information.

[0039] In addition, another embodiment further includes: Take any term as the target term, calculate the ratio between the sum of the occurrence frequencies of the target term in all legal provisions in the specific classification information and the total number of search results of the target term, and take the ratio as the relevant probability of the target term.

[0040] Specifically, the relevant probability satisfies the following relationship: ; In the formula, Indicates The entry is in The relevant probability of all search results of the class, Indicates The entry is in The number of class search results, Indicates The entry is in No. The frequency of occurrence of legal provisions, Indicates The total number of search results for terms.

[0041] Reflects the The entry is in The average frequency of occurrence in the class, that is, relative to the average of the total number of search results, the higher the relevant probability, the higher the probability of occurrence in the class. The entry is in The frequency of occurrence in the class is significantly higher than the average frequency in all search results, indicating that The entry is in The class has a stronger ability to distinguish and is more important; on the contrary, The ability to distinguish between classes is weak.

[0042] By passing the The entry is in The frequency of occurrence in the class is combined with the That is to say, it has been normalized and pays more attention to the absolute occurrence rate of the terms in specific classification information. It directly reflects the average occurrence rate of the terms in specific classification information and is independent of the distribution of the terms in other classification information. It is suitable for situations where there are a large number of classifications and a relatively uniform distribution.

[0043] Specifically, the importance weight of the entry satisfies the following relationship: ; In the formula, Indicates The importance weight of each term, Represents the maximum value function among all classification information, Indicates the total number of categories of legal provisions. Indicates The first The relevant probability of all search results of the class.

[0044] That is to say, if a term has a higher probability of being relevant in a certain category, then this term has a stronger ability to distinguish the category, and therefore should be given a higher weight. Therefore, the importance weight of a term can better reflect the maximum ability of the term to distinguish among all the category information, providing a more accurate basis for subsequent sorting and retrieval.

[0045] S3: Calculate the degree of correlation between any two terms in the related term set, obtain the term combination with higher frequency in the related term set, perform weakening processing based on the degree of correlation, obtain the weakening degree of each term combination and sum them up to obtain the overall weakening degree of the preliminary search results.

[0046] Relevance, including: Divide the number of times any two terms appear simultaneously in the preliminary search results by the total number of times the two terms appear in the preliminary search results respectively, and take the square root of the result. The square root is used as the correlation degree between the two terms.

[0047] Specifically, the correlation degree satisfies the following relationship: ; In the formula, Indicates Entries and The relevance of the terms, Indicates that the initial search results also include Entries and The number of search results for the term, and Respectively represent Entries and The total number of search results for terms.

[0048] It should be noted that in the process of calculating the priority by weighting the importance weight of the terms, there are cases where the correlation between terms is high and the importance weights are all high. It has strong classification ability and high importance weight, but the entry The appearance of is often accompanied by the term If they appear together, the terms will appear at the same time and entries The search results will be prioritized according to the terms and entries Duplicate entries are weighted to a greater extent, but in fact the strong classification capabilities of the two entries are the same, resulting in an artificially high priority. The priority of such entries needs to be adjusted.

[0049] In other words, only the terms that appear more frequently in the preliminary search results need to consider the weakening of the preliminary search results by the relevance degree, so the weakening degree is calculated according to the frequency of occurrence of each term in the terms.

[0050] The extent of the reduction includes: Calculate the joint occurrence frequency of any two terms in the preliminary search results. If the joint frequency is greater than the average occurrence frequency of all terms, the relevance of the two terms is weakened; otherwise, if the joint frequency is less than or equal to the average occurrence frequency of all terms, no weakening is required. The weakening method is: using a negative exponential function to weaken the relevance of two terms, so as to reduce the priority of the search results.

[0051] Specifically, the weakening degree satisfies the following relationship: ; In the formula, Indicates The first search result Entries and The degree of weakening of the term, Indicates Entries and The relevance of the terms, Indicates The first search result The frequency of occurrence of a term, Indicates The first search result The frequency of occurrence of a term, Indicates The average frequency of occurrence of all terms in the initial search results, Represented by natural numbers An exponential function with base .

[0052] That is to say, , explain Entries and The entry is in The joint occurrence frequencies in the preliminary search results are higher than the average and need to be weakened according to their relevance. Otherwise, they do not need to be weakened. Set to 1.

[0053] The overall weakening degree satisfies the following relationship: ; In the formula, Indicates The overall weakening of the initial search results, Represents the total number of terms in the related term set, Indicates The first search result Entries and The degree of weakening of an entry.

[0054] It should be noted that the outer loop from , inner loop from , ensuring that all possible combinations of terms are considered, and each term pair is counted only once; by index The cumulative multiplication results are adjusted and averaged, which is the inverse of the number of all term combinations, and is used to normalize the cumulative multiplication results to obtain a reasonable average weakening degree.

[0055] S4: perform a weighted summation of the importance weight and the frequency of occurrence of the term in the legal text, and make corrections based on the overall weakening degree to obtain the priority of the final search results, sort the final search results according to the priority, and obtain the final display results.

[0056] Specifically, the priority satisfies the following relationship: ; In the formula, Indicates The priority of a legal provision, Indicates The overall weakening of the initial search results, Represents the total number of terms in the related term set, Indicates The importance weight of each term, Indicates The final search results The frequency of occurrence of a term.

[0057] In other words, the priority The higher the value, the stronger the relevance of the legal provision to the user's search intent. The higher the frequency of its appearance in legal texts The higher the number, the higher the priority of the legal provision.

[0058] The final search results are sorted from largest to smallest based on priority and output in a list format for viewing the most relevant legal provisions.

[0059] That is to say, all legal provisions are sorted from high to low according to their priority. A higher priority means that the legal provision is more compatible with the user's search intent and should be ranked first. On the contrary, a lower priority means that the legal provision is less relevant and should be ranked last.

[0060] It should be noted that the search results are sorted according to their priority to ensure that users can quickly find the most relevant and most appropriate legal provisions. The priority is calculated based on the importance weight of the terms and their frequency of occurrence in the legal provisions, so the sorting results can reflect the relevance and accuracy of the search results.

[0061] The present invention also provides a legal text retrieval system. Figure 2 As shown, the system includes a processor and a memory, the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a legal text retrieval method according to the first aspect of the present invention is implemented. The system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, whose settings and functions are known in the art, and therefore will not be described in detail here.

[0062] It should be noted that, for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these modifications and improvements all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A legal text retrieval method, characterized in that: include: Obtaining a set of related terms in the Boolean model, performing a database search on the terms in the set of related terms, obtaining a preliminary search result containing any term in the set of related terms, and screening out a search result that strictly matches the Boolean expression query condition from the preliminary search result as a final search result; Obtaining classification information of the terms, counting the frequency of occurrence of each term in each classification information, calculating the relevant probability of the term in each classification information, and selecting the largest relevant probability in each classification information as the importance weight of the term, wherein the classification information refers to various classification attributes related to the legal provisions; Calculate the correlation between any two terms in the related term set, obtain the term combination with a higher frequency in the related term set, perform weakening processing based on the correlation, obtain the weakening degree of each term combination and sum them up to obtain the overall weakening degree of the preliminary search results; The importance weight and the frequency of occurrence of the term in the legal text are weighted and summed, and corrections are made based on the overall weakening degree to obtain the priority of the final search results. The final search results are sorted according to the priority to obtain the final display results.

2. A legal text retrieval method according to claim 1, characterized in that: The preliminary search results are all legal provisions containing relevant terms retrieved based on the Boolean model, and the final search results are legal provisions that fully meet the user's search conditions screened out from the preliminary search results; In response to the legal text containing any term in the Boolean model, the legal text is considered to be a preliminary search result. In response to the legal text containing all terms in the Boolean model, the legal text is considered to be a final detection result.

3. A legal text retrieval method according to claim 1, characterized in that: The relevant probabilities include: Taking any term as the target term, calculate the ratio between the sum of the frequency of occurrence of the target term in all legal articles in a specific classification and the sum of the frequency of occurrence of all legal articles in all classifications, and obtain the relative importance of the target term in the specific classification; Calculate the ratio of the number of legal articles of the target term in the specific classification information to the total number of search results to obtain the concentration of the target term in the specific classification information; The product of the relative importance and the concentration level is used as the relevance probability of the target term in the specific classification information.

4. A legal text retrieval method according to claim 1, characterized in that: The related probability also includes: Taking any term as the target term, the ratio between the sum of the occurrence frequencies of the target term in all legal provisions in the specific classification information and the total number of search results of the target term is calculated, and the ratio is taken as the relevant probability of the target term.

5. A legal text retrieval method according to claim 1, characterized in that: The degree of relevance includes: Divide the number of times any two terms appear simultaneously in the preliminary search results by the total number of times the two terms appear in the preliminary search results respectively, and take the square root of the result. The square root is used as the correlation degree between the two terms.

6. A legal text retrieval method according to claim 1, characterized in that: The degree of weakening includes: Calculate the joint occurrence frequency of any two terms in the preliminary search results. If the joint frequency is greater than the average occurrence frequency of all terms, the relevance of the two terms is weakened; otherwise, if the joint frequency is less than or equal to the average occurrence frequency of all terms, no weakening is required. The weakening method is: using a negative exponential function to weaken the relevance of two terms, so as to reduce the priority of the search results.

7. A legal text retrieval method according to claim 6, characterized in that: The overall weakening degree satisfies the following relationship: ； In the formula, Indicates The overall weakening of the initial search results, Represents the total number of terms in the related term set, Indicates The first search result Entries and The degree of weakening of an entry.

8. A legal text retrieval method according to claim 1, characterized in that: The priority satisfies the following relationship: ； In the formula, Indicates The priority of the initial search results, Indicates The overall weakening of the initial search results, Represents the total number of terms in the related term set, Indicates The importance weight of each term, Indicates The final search results are selected from the preliminary search results. The frequency of occurrence of a term.

9. A legal text retrieval method according to claim 1, characterized in that: The obtaining of the final display result includes: The final search results are sorted from largest to smallest based on priority and output in a list format for viewing the most relevant legal provisions.

10. A legal text retrieval system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the legal text retrieval method according to any one of claims 1 to 9 is implemented.

Images