Case series-parallel method and device based on natural language processing technology and electronic equipment

Abstract

The invention relates to a case series-parallel method and device based on a natural language processing technology and electronic equipment, and belongs to the technical field of natural language process.The method comprises the steps that key element entities of historical cases are obtained; performing feature vectorization on the key element entities of the historical cases to obtain a first feature vector of each historical case; obtaining a second feature vector of the to-be-processed case; calculating the similarity between the second feature vector and the first feature vector to obtain a first similarity value; calculating the similarity between a set formed by the key element entities of each historical case and a set formed by the key element entities of the to-be-processed case to obtain a second similarity value; and on the basis of the first similarity value and the second similarity value, determining a series-parallel case of the to-be-processed case and the historical case. According to the method, the key element entities of the cases are automatically extracted through the natural language, and then the series-parallel cases related to the cases are determined through similarity calculation, so that the case series-parallel efficiency in a judicial system is improved.

Description

Technical Field

[0001] The present invention relates to the field of natural language processing technology, and in particular to a case concatenation method, device and electronic equipment based on natural language processing technology. Background Art

[0002] In the handling of judicial cases, especially in criminal cases and complex economic crime cases, a large number of case records are usually involved. These records record in detail the facts of the case, the confessions of the persons involved, the testimony of witnesses and other relevant information. In actual operations, there may be certain correlations between different cases. For example, multiple cases may involve specific targets, similar methods of committing crimes, regular crime times or the same persons involved. This correlation has brought the judicial authorities the need to handle cases in series (analyze multiple similar cases in series, compare and analyze the commonalities and differences between them to support theoretical research and trend analysis, and provide a direction for solving the case) and merge cases (unify and analyze related cases, and the objects of analysis may be the same criminal subject, so as to handle them efficiently and solve the problem as a whole).

[0003] Traditional case analysis mainly relies on manual reading and empirical judgment, but with the surge in the number of cases, this method can no longer meet the needs of the modern judicial system due to its low efficiency. Summary of the invention

[0004] In view of this, it is necessary to provide a case concatenation method, device and electronic device based on natural language processing technology to solve the problem of low efficiency caused by manual concatenation of cases in the existing judicial system.

[0005] In order to solve the above problems, the present invention provides a case concatenation method based on natural language processing technology, comprising: Obtain key entity elements of historical cases; Perform feature vectorization on the key element entities of historical cases to obtain the first feature vector of each historical case; Obtain the second eigenvector of the case to be processed; Calculating the similarity between the second eigenvector and the first eigenvector to obtain a first similarity value; Calculate the similarity between the set of key element entities of each historical case and the set of key element entities of the case to be processed to obtain a second similarity value; Based on the first similarity value and the second similarity value, it is determined whether the case to be processed and the historical case are serial cases or combined cases.

[0006] In a possible implementation, the step of obtaining key element entities of historical cases includes: Preprocessing the transcript data of historical cases to obtain preprocessed transcript data; Input the preprocessed transcript data into a preset named entity recognition model to extract case features and obtain specific types of entities; The key element entities are screened out from the entities of the specific type.

[0007] In a possible implementation, the specific type of entity includes: one or more of the following: crime scene characteristics, crime time characteristics, crime address characteristics, crime target characteristics, escape method characteristics, suspect characteristics, and victim characteristics; The key element entities include: one or more of: mobile phone number, ID number, license plate number, institution location, bank card number, WeChat number, QQ number, QQ group number and Alipay account.

[0008] In a possible implementation, calculating the similarity between the second feature vector and the first feature vector to obtain a first similarity value includes: The cosine similarity between the second eigenvector and the first eigenvector is calculated to obtain a first similarity value.

[0009] In a possible implementation, the calculating the similarity between the set of key element entities of each historical case and the set of key element entities of the case to be processed to obtain the second similarity value includes: The Jaccard similarity coefficient is calculated between the set of key element entities of each historical case and the set of key element entities of the case to be processed to obtain a second similarity value.

[0010] In a possible implementation, the method further includes: The alarm information corresponding to the key element entity of the historical case and the key element entity of the historical case are stored in the database; Determine historical cases related to the case to be processed based on the key element entities of the case to be processed and the database; Based on the related historical cases, determine whether to-be-processed cases should be linked or merged with historical cases.

[0011] In a possible implementation, determining the historical cases associated with the case to be processed based on the key element entity of the case to be processed and the database includes: The key element entities of the case to be processed are used as keywords, and group queries are performed in the database based on the keywords to determine historical cases related to the case to be processed.

[0012] On the other hand, the present invention also provides a case stringing and paralleling device based on natural language processing technology, comprising: A historical case processing module is used to obtain key element entities of historical cases; A first feature vector acquisition module, used to perform feature vectorization on key element entities of historical cases to obtain a first feature vector for each historical case; A second feature vector acquisition module, used to acquire a second feature vector of the case to be processed; A first similarity value acquisition module, used to calculate the similarity between the second feature vector and the first feature vector to obtain a first similarity value; A second similarity value acquisition module is used to calculate the similarity between the set of key element entities of each historical case and the set of key element entities of the case to be processed, so as to obtain a second similarity value; A module for determining whether a case to be processed and a historical case are connected or combined based on the first similarity value and the second similarity value.

[0013] On the other hand, the present invention also provides an electronic device, including a memory and a processor, wherein: The memory is used to store programs; The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in a case stringing and paralleling method based on natural language processing technology described in any one of the above implementation methods.

[0014] On the other hand, the present invention also provides a computer-readable storage medium for storing computer-readable programs or instructions, which, when executed by a processor, can implement the steps of a case stringing and paralleling method based on natural language processing technology described in any of the above-mentioned implementation methods.

[0015] The beneficial effects of the present invention are as follows: the present invention provides a case stringing and merging method based on natural language processing technology, firstly obtains the key element entities of historical cases, performs feature vectorization on the key element entities of historical cases, obtains the first feature vector of each historical case, so that the case to be processed can calculate the similarity with the historical case, then obtains the second feature vector of the case to be processed, calculates the similarity between the second feature vector and the first feature vector, obtains the first similarity value, calculates the similarity between the set composed of the key element entities of each historical case and the set composed of the key element entities of the case to be processed, obtains the second similarity value, and determines the stringing or merging of the case to be processed with the historical case based on the first similarity value and the second similarity value. The present invention automatically extracts the key element entities of the case through natural language, and then determines the stringing and merging cases related to the case through similarity calculation, thereby improving the efficiency of stringing and merging cases in the judicial system. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flow chart of a structure and method of an embodiment of a case stringing and paralleling method based on natural language processing technology provided by the present invention; Figure 2 For the present invention Figure 1 A schematic flow chart of an embodiment of S101; Figure 3 A schematic flow chart of an embodiment of a case serial and parallel device based on natural language processing technology provided by the present invention; Figure 4 A schematic structural diagram of an embodiment of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0017] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0018] Before presenting the embodiments, the following terms are explained.

[0019] A series of different cases are solved together by analyzing the methods of committing the crime, traces, physical evidence, etc. and finding that they are related.

[0020] A joint case means two different cases that are connected through the methods of committing the crime, traces, and physical evidence, so the two cases are solved together.

[0021] The present invention provides a case stringing and paralleling method, device and electronic device based on natural language processing technology, which are described below respectively.

[0022] Figure 1 A flow chart of an embodiment of the case stringing and paralleling method based on natural language processing technology provided by the present invention is as follows: Figure 1 As shown in the figure, the case stringing and merging method based on natural language processing technology includes: S101. Obtain key element entities of historical cases; S102, quantizing the key element entities of the historical cases to obtain the first feature vector of each historical case; S103, obtaining a second feature vector of the case to be processed; S104, calculating the similarity between the second eigenvector and the first eigenvector to obtain a first similarity value; S105, calculating the similarity between the set of key element entities of each historical case and the set of key element entities of the case to be processed, to obtain a second similarity value; S106. Determine whether the case to be processed and the historical case are serial cases or combined cases based on the first similarity value and the second similarity value.

[0023] Compared with the prior art, the present embodiment provides a case stringing and merging method based on natural language processing technology, firstly obtains the key element entities of historical cases, performs feature vectorization on the key element entities of historical cases, obtains the first feature vector of each historical case, so that the case to be processed can be calculated similar to the historical case, then obtains the second feature vector of the case to be processed, calculates the similarity between the second feature vector and the first feature vector, obtains the first similarity value, calculates the similarity between the set composed of the key element entities of each historical case and the set composed of the key element entities of the case to be processed, obtains the second similarity value, and determines the stringing or merging of the case to be processed with the historical case based on the first similarity value and the second similarity value. The present invention automatically extracts the key element entities of the case through natural language, and then determines the stringing and merging cases related to the case through similarity calculation, thereby improving the efficiency of stringing and merging cases in the judicial system.

[0024] In some embodiments of the present invention, in step S101, Figure 2 As shown, the key element entities of the historical case acquisition include: S201, preprocessing the transcript data of historical cases to obtain preprocessed transcript data; S202, inputting the preprocessed transcript data into a preset named entity recognition model to extract case features to obtain specific type entities; S203: Filter out key element entities from the specific type of entities.

[0025] In a specific embodiment of the present invention, in step S201, the transcript data of all historical cases are collected and preprocessed, including text cleaning and format standardization. Text cleaning includes removing unnecessary special symbols, punctuation marks and spaces, correcting spelling errors, and standardizing the wording of case transcripts. Format standardization includes unifying the formats of transcript data from different sources to ensure consistency in subsequent processing.

[0026] Examples of irregular transcripts (partial) Zhang said: "On the evening of XX month XX day, I was strolling on XX Road and saw that there was only one clerk in the convenience store. I entered the store at about XX evening, put on a hat and a mask, took out dangerous tools, and threatened the clerk to hand over the money. The clerk was quite scared, so he gave me the money in the cash register, about 500 yuan. I took the money and went home." Example of standardized transcript (partial) The person being questioned, Zhang XX, stated: “At around XX:00 on ... In a specific embodiment of the present invention, NLP technology is used to process and analyze the transcript text to extract case features and key element entities. Case feature extraction mainly includes word segmentation and part-of-speech tagging, and training the Named Entity Recognition (NER) model to extract specific types of entities (these entities are usually important features of case analysis). Key element entity extraction mainly extracts identity recognition features from the case feature set as key element entities, and also needs to identify the attribution type of the element entity.

[0027] Word segmentation and part-of-speech tagging include word segmentation of the transcript text and tagging the part-of-speech, for example: "Zhang robbed a convenience store of 500 yuan in cash on the night of XX month XX day of XXXX." After word segmentation and part-of-speech tagging, we can know that: Zhang (noun: suspect) robbed (verb: means of committing the crime) on the night of (preposition) XX month XX day of XXXX (noun: time of the crime) (particle) a (numeral + quantifier) ​​convenience store (name: location of the crime) 500 (numeral) yuan (quantifier) ​​in cash (noun: target of the crime). (punctuation marks) Train a suitable named entity recognition model. First, use the labeled data to train the model so that the model can better identify specific types of entities in the transcript content. Then use the test set to evaluate the trained model, such as accuracy and recall, and then optimize the model parameters and increase the amount of training based on the evaluation results to improve the model effect. Finally, use the trained model to extract the pre-processed historical transcript text.

[0028] In some embodiments of the present invention, the specific type of entity includes: one or more of the following: crime scene characteristics, crime time characteristics, crime address characteristics, crime target characteristics, escape method characteristics, suspect characteristics, and victim characteristics; The key element entities include: one or more of: mobile phone number, ID number, license plate number, institution location, bank card number, WeChat number, QQ number, QQ group number and Alipay account.

[0029] The specific types of entities that may be extracted by the named entity recognition model generally include crime scene characteristics (specific marks, destructive behavior), crime time characteristics (specific time period, day of the week, season), crime address characteristics (latitude and longitude, location type), crime target characteristics (property type, specific item selection), crime method characteristics (weapon type, degree of violence), escape method characteristics (transportation, escape direction), suspect characteristics (gender, age, height, body shape, mobile phone number, online account), and victim characteristics (gender, age, occupation, mobile phone number, online account).

[0030] The key element entities are mainly to filter out identity identification features (unique identifiers) from the above features, such as mobile phone number, ID number, license plate number, institution location, bank card number, WeChat number, QQ number, QQ group number, Alipay account, etc.

[0031] At the same time, it is also necessary to identify the relationship between key element entities or the type of element attribution. For example, after the features such as "Yuehuoli Supermarket, Shucheng Road, Hongshan District, Wuhan City", "mobile phone number", and "18171555555" have been identified in a certain transcript, it is necessary to identify that "18171555555" belongs to "mobile phone number" and "Yuehuoli Supermarket, Shucheng Road, Hongshan District, Wuhan City" belongs to "institutional place" (there is no "institutional place" in the identified features, and semantic analysis is required to perform element type attribution analysis).

[0032] In a specific embodiment of the present invention, the features generally used for case analysis need to give priority to features directly related to criminal behavior and crime patterns, such as crime means features, crime target features, crime time features, etc. Secondly, features that are relatively stable and recurring in different cases, such as suspect features, victim features, etc., are selected.

[0033] Some of the above features can be directly extracted in step S102, and some of them need to be constructed, such as extracting the year, month, day, hour, week, season, etc. from the directly extracted crime time to construct the crime time feature.

[0034] The above text features are vectorized, generally using methods such as Bag of Words, TF-IDF, One-Hot Encoding, etc., and then the obtained numerical vectors are standardized (Z-score) and normalized (Min-Max) to make different features have similar scales and distributions, which is convenient for subsequent calculations and analysis.

[0035] Bag of Words Model: First, the text is tokenized, and then the number of times each word appears in the text is counted to form a term frequency vector. For example, for the text "I put on a hat and a mask", after being processed by the bag of words model, a vector [0, 1, 1, 0] may be obtained. Assuming the dictionary is ["shoes", "hat", "mask", "glasses"], where "hat" and "mask" appear, so the corresponding positions are 1.

[0036] Term Frequency - Inverse Document Frequency: Combines term frequency (TF) and inverse document frequency (IDF) to measure the importance of a word. TF represents the frequency of a word in the current text (number of times the word appears / total number of words in the text), and IDF represents the prevalence of a word in the entire document collection. The IDF of rare words is higher (total number of documents / number of documents in which the word appears, and then take the natural logarithm of the result). For example, the WeChat ID "Xiaoi 9527" appears less frequently in other transcripts, so its IDF value is higher, and then the vector of this text may highlight the importance of the word "Xiaoi 9527".

[0037] Standardization: Calculates the mean and standard deviation of the features. For each feature value x, it is standardized using the formula (x - mean) / standard deviation. For example, for a set of term frequency features [10, 20, 30, 40, 50], the mean is 30 and the standard deviation is 14.14, then the standardized feature vector is [(-1.41), (-0.71), (0), (0.71), (1.41)].

[0038] Normalization: First determines the minimum and maximum values of the features. For each feature value x, it is normalized using the formula (x - minimum value) / (maximum value - minimum value). For example, for the same term frequency features above, the minimum value is 10 and the maximum value is 50, then the normalized features are [0, 0.25, 0.5, 0.75, 1].

[0039] In some embodiments of the present invention, calculating the similarity between the second feature vector and the first feature vector to obtain a first similarity value includes: Calculating the cosine similarity between the second feature vector and the first feature vector to obtain a first similarity value.

[0040] In some embodiments of the present invention, calculating the similarity between the set composed of the key element entities of each historical case and the set composed of the key element entities of the case to be processed to obtain a second similarity value includes: Calculating the Jaccard similarity coefficient between the set composed of the key element entities of each historical case and the set composed of the key element entities of the case to be processed to obtain a second similarity value.

[0041] In the specific implementation of the present invention, the similarity between cases is calculated to identify possible serial cases and combined cases. Common methods for calculating similarity in mathematics include cosine similarity and Jaccard similarity coefficient.

[0042] Cosine Similarity: Calculates the cosine similarity between case feature vectors.

[0043] For two non-zero vectors A and B, their cosine similarity is defined as the dot product of the two vectors divided by the product of their moduli, that is: cosine_similarity(A, B) = (A · B) / (||A|| · ||B||) Where A · B represents the dot product of vectors A and B, and ||A|| and ||B|| represent the modulus of vectors A and B respectively.

[0044] In practical applications, such as text processing, the text is first converted into a numerical vector (for example, through the bag-of-words model, TF-IDF, etc.), and then the cosine similarity between these vectors is calculated.

[0045] The value range of cosine similarity is between -1 and 1: 1 means that the two vectors are exactly the same and have the same direction.

[0046] 0 means the two vectors are perpendicular to each other and have no similarity.

[0047] -1 means the two vectors are completely opposite.

[0048] For example, suppose there are two text vectors A = [1, 2, 3] and B = [2, 4, 6], A · B = 1·2 + 2·4 + 3·6 = 28 ||A|| = √(1^2 + 2^2 + 3^2) = √14 ||B|| = √(2^2 + 4^2 + 6^2) = √56 Then their cosine similarity is: cosine_similarity(A, B) = 28 / (√14 · √56) ≈ 1 , indicating that the two texts are very similar.

[0049] Jaccard similarity coefficient: Calculates the similarity between cases based on a set of keywords.

[0050] For two sets A and B, the Jaccard similarity coefficient is defined as the number of elements in the intersection of the two sets divided by the number of elements in the union, that is: Jaccard(A, B) = |A ∩ B| / |A ∪B| Among them, |A ∩ B| represents the number of elements in the intersection of sets A and B, and |A ∪B| represents the number of elements in the union of sets A and B.

[0051] The value range of Jaccard similarity coefficient is between [0, 1]: 0 means that the two sets have no common elements and are completely dissimilar.

[0052] 1 means the two sets are identical.

[0053] For example, set A = {1, 2, 3, 4, 5}, set B = {3, 4, 5, 6, 7} Then A ∩ B = {3, 4, 5}, |A ∩ B| = 3 A ∪B = {1, 2, 3, 4, 5, 6, 7}, |A ∪B| = 7 The Jaccard similarity coefficient Jaccard(A, B) = 3 / 7 ≈ 0.43, indicating that the two sets have a certain similarity, but not very high.

[0054] If the cosine similarity calculation results of the key element word vectors of the transcripts of two cases and the Jaccard similarity coefficient calculation results of their key element sets are both close to 1, it means that the similarity between the two cases is very high, otherwise the similarity is low.

[0055] In some embodiments of the present invention, it further includes: The alarm information corresponding to the key element entity of the historical case and the key element entity of the historical case are stored in the database; Determine historical cases related to the case to be processed based on the key element entities of the case to be processed and the database; Based on the related historical cases, determine whether to-be-processed cases should be linked or merged with historical cases.

[0056] In some embodiments of the present invention, determining the historical cases associated with the case to be processed based on the key element entities of the case to be processed and the database includes: The key element entities of the case to be processed are used as keywords, and group queries are performed in the database based on the keywords to determine historical cases related to the case to be processed.

[0057] In a specific embodiment of the present invention, similarity calculation is performed between the cases to be processed and the historical cases to determine the serial and combined cases of the cases to be processed. In order to further increase the scope of the serial and combined cases with the cases to be processed, the key element entities of the historical cases and the corresponding police information are put into the database. When there are new cases to be processed, the related cases of the new cases to be processed are determined through group query statements, that is, the serial and combined cases related to the new cases to be processed.

[0058] Specifically, the key element entities extracted from a certain transcript are grouped to ensure that there is only one element in each group. For example, a certain transcript extracts five key element entities, namely, mobile phone number, ID number, bank card number, WeChat number, and Alipay account. The transcript number, transcript name, case number, case name, alarm number, alarm name, and incident time corresponding to each transcript are combined and written into the transcript element index library (all transcripts must be processed as above). Then, the number of times each element is associated with a transcript, associated case, or associated police situation, as well as the earliest and latest collection time of the element, are counted through a scheduled task. The statistical result table is stored separately in a relational database. Based on this table, the association relationship between the key element entity of the case to be added and the historical transcript, case, and police situation, the number of associations, the earliest association time, the latest association time, etc., can be queried, so as to determine the serial and parallel cases related to the newly added cases to be processed.

[0059] In order to better implement a case stringing and combining method based on natural language processing technology in an embodiment of the present invention, based on a case stringing and combining method based on natural language processing technology, correspondingly, as follows Figure 3 As shown, the embodiment of the present invention further provides a case stringing and paralleling device based on natural language processing technology. The case stringing and paralleling device 300 based on natural language processing technology includes: A historical case processing module 301 is used to obtain key element entities of historical cases; A first feature vector acquisition module 302 is used to perform feature vectorization on key element entities of historical cases to obtain a first feature vector for each historical case; A second feature vector acquisition module 303, used to acquire a second feature vector of the case to be processed; A first similarity value acquisition module 304 is used to calculate the similarity between the second feature vector and the first feature vector to obtain a first similarity value; A second similarity value acquisition module 305 is used to calculate the similarity between the set of key element entities of each historical case and the set of key element entities of the case to be processed, and obtain a second similarity value; The module 306 for determining whether a case to be processed and a historical case are connected or combined is used to determine whether the case to be processed and the historical case are connected or combined based on the first similarity value and the second similarity value.

[0060] The case stringing and paralleling device 300 based on natural language processing technology provided in the above embodiment can implement the technical solution described in the above embodiment of a case stringing and paralleling method based on natural language processing technology. The specific implementation principles of the above modules or units can refer to the corresponding contents in the above embodiment of a case stringing and paralleling method based on natural language processing technology, which will not be repeated here.

[0061] like Figure 4 As shown, the present invention also provides an electronic device 400. The electronic device 400 includes a processor 401, a memory 402 and a display 403. Figure 4 Only some components of the electronic device 400 are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0062] In some embodiments, the processor 401 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run program codes or process data stored in the memory 402, such as a method for coordinated speed control in a diversion section of a highway in the present invention.

[0063] In some embodiments, processor 401 may be a single server or a server group. The server group may be centralized or distributed. In some embodiments, processor 401 may be local or remote. In some embodiments, processor 401 may be implemented in a cloud platform. In some embodiments, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-cloud, etc., or any combination thereof.

[0064] In some embodiments, the memory 402 may be an internal storage unit of the electronic device 400, such as a hard disk or memory of the electronic device 400. In other embodiments, the memory 402 may also be an external storage device of the electronic device 400, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc., equipped on the electronic device 400.

[0065] Furthermore, the memory 402 may include both an internal storage unit of the electronic device 400 and an external storage device. The memory 402 is used to store application software installed in the electronic device 400 and various data.

[0066] In some embodiments, the display 403 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 403 is used to display information on the electronic device 400 and to display a visual user interface. The components 401-403 of the electronic device 400 communicate with each other via a system bus.

[0067] In one embodiment, when the processor 401 executes a case stringing and paralleling program based on natural language processing technology in the memory 402, the following steps may be implemented: Obtain key entity elements of historical cases; Perform feature vectorization on the key element entities of historical cases to obtain the first feature vector of each historical case; Obtain the second feature vector of the case to be processed; Calculating the similarity between the second eigenvector and the first eigenvector to obtain a first similarity value; Calculate the similarity between the set of key element entities of each historical case and the set of key element entities of the case to be processed to obtain a second similarity value; Based on the first similarity value and the second similarity value, it is determined that the case to be processed and the historical case are serial cases or combined cases. It should be understood that when the processor 401 executes a case serialization and combination program based on natural language processing technology in the memory 402, in addition to the above functions, other functions can also be realized. For details, please refer to the description of the corresponding method embodiment above.

[0068] Furthermore, the embodiment of the present invention does not specifically limit the type of the electronic device 400 mentioned, and the electronic device 400 may be a portable electronic device such as a mobile phone, a tablet computer, a personal digital assistant (PDA), a wearable device, a laptop computer, etc. Exemplary embodiments of portable electronic devices include but are not limited to portable electronic devices equipped with IOS, Android, Microsoft or other operating systems. The above-mentioned portable electronic devices may also be other portable electronic devices, such as a laptop computer with a touch-sensitive surface (e.g., a touch panel). It should also be understood that in some other embodiments of the present invention, the electronic device 400 may not be a portable electronic device, but a desktop computer with a touch-sensitive surface (e.g., a touch panel).

[0069] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0070] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A case concatenation method based on natural language processing technology, characterized in that: include: Obtain key entity elements of historical cases; Perform feature vectorization on the key element entities of historical cases to obtain the first feature vector of each historical case; Obtain the second feature vector of the case to be processed; Calculating the similarity between the second eigenvector and the first eigenvector to obtain a first similarity value; Calculate the similarity between the set of key element entities of each historical case and the set of key element entities of the case to be processed to obtain a second similarity value; Based on the first similarity value and the second similarity value, it is determined whether the case to be processed and the historical case are serial cases or combined cases.

2. The case concatenation and parallelization method based on natural language processing technology according to claim 1 is characterized in that: The key element entities of obtaining historical cases include: Preprocessing the transcript data of historical cases to obtain preprocessed transcript data; Input the preprocessed transcript data into a preset named entity recognition model to extract case features and obtain specific types of entities; The key element entities are screened out from the entities of the specific type.

3. The case concatenation and parallelization method based on natural language processing technology according to claim 2 is characterized in that: The specific type of entity includes: one or more of the following: crime scene characteristics, crime time characteristics, crime address characteristics, crime target characteristics, escape method characteristics, suspect characteristics, and victim characteristics; The key element entities include: one or more of: mobile phone number, ID number, license plate number, institution location, bank card number, WeChat number, QQ number, QQ group number and Alipay account.

4. The case concatenation and parallelization method based on natural language processing technology according to claim 1 is characterized in that: The calculating the similarity between the second feature vector and the first feature vector to obtain a first similarity value includes: The cosine similarity between the second eigenvector and the first eigenvector is calculated to obtain a first similarity value.

5. The case concatenation and parallelization method based on natural language processing technology according to claim 1 is characterized in that: The calculating the similarity between the set of key element entities of each historical case and the set of key element entities of the case to be processed to obtain a second similarity value includes: The Jaccard similarity coefficient is calculated between the set of key element entities of each historical case and the set of key element entities of the case to be processed to obtain a second similarity value.

6. The case concatenation and parallelization method based on natural language processing technology according to claim 1 is characterized in that: Also includes: The alarm information corresponding to the key element entity of the historical case and the key element entity of the historical case are stored in the database; Determine historical cases related to the case to be processed based on the key element entities of the case to be processed and the database; Based on the related historical cases, determine whether to-be-processed cases should be linked or merged with historical cases.

7. The case concatenation and parallelization method based on natural language processing technology according to claim 6 is characterized in that: The determining of historical cases associated with the case to be processed based on the key element entity of the case to be processed and the database includes: The key element entities of the case to be processed are used as keywords, and group queries are performed in the database based on the keywords to determine historical cases related to the case to be processed.

8. A case stringing and paralleling device based on natural language processing technology, characterized in that: include: A historical case processing module is used to obtain key element entities of historical cases; A first feature vector acquisition module, used to perform feature vectorization on key element entities of historical cases to obtain a first feature vector for each historical case; A second feature vector acquisition module, used to acquire a second feature vector of the case to be processed; A first similarity value acquisition module, used to calculate the similarity between the second feature vector and the first feature vector to obtain a first similarity value; A second similarity value acquisition module is used to calculate the similarity between the set of key element entities of each historical case and the set of key element entities of the case to be processed, and obtain a second similarity value; A module for determining whether a case to be processed and a historical case are connected or combined based on the first similarity value and the second similarity value.

9. An electronic device, characterized in that: comprising a memory and a processor, wherein: The memory is used to store programs; The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the case stringing and paralleling method based on natural language processing technology as described in any one of claims 1 to 7 above.

10. A computer-readable storage medium, characterized in that: Used to store computer-readable programs or instructions, which, when executed by a processor, can implement the steps in a case stringing and merging method based on natural language processing technology as described in any one of claims 1 to 7 above.

Images