Case merging method, device, and storage medium

Abstract

This application discloses a case merging method, apparatus, and storage medium. The case merging method includes: extracting elements from the descriptive information of the current case to obtain several initial elements of the current case; fusing the initial elements of the current case with historical elements of several historical cases to obtain target elements of the current case and each historical case; and merging the current case and the several historical cases based on the similarity between the target elements of the current case and the target elements of the several historical cases. This approach can improve the efficiency of case merging.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus and storage medium for merging cases. Background Technology

[0002] Case merging has always been a challenging problem in data processing. Currently, it mainly relies on manual data analysis to merge individual cases. However, as the number of cases increases, this manual data analysis method becomes inefficient and unsuitable for large-scale cases. Therefore, there is an urgent need for a method to improve the efficiency of case merging. Summary of the Invention

[0003] This application provides at least one method, apparatus, and storage medium for merging cases.

[0004] This application provides a case merging method, including: extracting elements from the descriptive information of the current case to obtain several initial elements of the current case; merging the initial elements of the current case with historical elements of several historical cases to obtain target elements of the current case and each historical case; and merging the current case and several historical cases based on the similarity between the target elements of the current case and the target elements of several historical cases.

[0005] This application provides a case merging device, comprising: an element extraction module for extracting elements from the descriptive information of the current case to obtain several initial elements of the current case; an element merging module for fusing the initial elements of the current case with historical elements of several historical cases to obtain target elements of the current case and each historical case; and a case merging module for merging the current case and several historical cases based on the similarity between the target elements of the current case and the target elements of several historical cases.

[0006] This application provides an electronic device, including a memory and a processor, wherein the processor is used to execute program instructions stored in the memory to implement the above-described case merging method.

[0007] This application provides a computer-readable storage medium storing program instructions thereon, which, when executed by a processor, implement the above-described case merging method.

[0008] The above-described solution extracts elements from the descriptive information of the current case, then merges the current case and several historical cases based on the similarity between the elements of the current case and the elements of several historical cases. This eliminates the need for manual data analysis, improving the efficiency of case merging. Furthermore, by fusing the initial elements of the current case with the historical elements of several historical cases to obtain the target elements of both the current case and each historical case, and then merging the current case and several historical cases based on the similarity between the target elements of the current case and each historical case, the accuracy of case merging is further enhanced.

[0009] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this application. Attached Figure Description

[0010] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with this application and, together with the specification, serve to explain the technical solutions of this application.

[0011] Figure 1 This is a flowchart illustrating one embodiment of the method for merging cases in this application;

[0012] Figure 2 This is a schematic diagram of the structure of an embodiment of the case merging device of this application;

[0013] Figure 3 This is a schematic diagram of the structure of an embodiment of the electronic device of this application;

[0014] Figure 4 This is a schematic diagram of the structure of an embodiment of the computer-readable storage medium of this application. Detailed Implementation

[0015] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0016] In the following description, specific details such as particular system architectures, interfaces, and technologies are presented for illustrative purposes rather than for limiting purposes, in order to provide a thorough understanding of this application.

[0017] In this document, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship. Furthermore, "many" in this document means two or more. Moreover, the term "at least one" in this document means any combination of at least two of any one or more of a plurality of objects. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0018] Please see Figure 1 , Figure 1 This is a flowchart illustrating one embodiment of the method for merging cases in this application. Figure 1 As shown, the case merging method provided in this disclosure embodiment may include the following steps:

[0019] Step S11: Extract elements from the descriptive information of the current case to obtain several initial elements of the current case.

[0020] The "several" mentioned in the embodiments of this disclosure refers to one or more. The initial element may be an event element related to the current case. For example, an event element may include the object of the action, the location of the occurrence, the time of the occurrence, the subject of the action, the action, etc.

[0021] Step S12: Integrate the initial elements of the current case with the historical elements of several historical cases to obtain the target elements of the current case and each historical case.

[0022] In some disclosed embodiments, if only one historical case exists prior to the current case, the historical elements of that historical case are its initial elements. The initial elements of the current case and the initial elements of the historical case are then merged to obtain the target elements of the current case and the target elements of the historical case. The current case and the historical case can serve as the historical cases for the next case, and the target elements of the current case and the historical case serve as historical elements. The merging of initial elements and historical elements can be achieved by merging identical elements from different cases.

[0023] Step S13: Based on the similarity between the target elements of the current case and the target elements of several historical cases, merge the current case and several historical cases.

[0024] One method to determine the similarity between the target elements of the current case and the target elements of several historical cases is to calculate the distance between each target element and determine the similarity based on the length of the distance. Cases that meet the similarity criteria can be grouped together, while cases that do not meet the similarity criteria will not be grouped together.

[0025] The above-described solution extracts elements from the descriptive information of the current case, then merges the current case and several historical cases based on the similarity between the elements of the current case and the elements of several historical cases. This eliminates the need for manual data analysis, improving the efficiency of case merging. Furthermore, by fusing the initial elements of the current case with the historical elements of several historical cases to obtain the target elements of both the current case and each historical case, and then merging the current case and several historical cases based on the similarity between the target elements of the current case and each historical case, the accuracy of case merging is further enhanced.

[0026] In some disclosed embodiments, before extracting elements, the descriptive information of the current case is used to extract the case type. This can be done using text classification methods in natural language processing. Optionally, before classifying the current case using its descriptive information to obtain its case type, the descriptive information can be preprocessed. Preprocessing methods can include word segmentation, stop word removal, etc. In some disclosed embodiments, different case types are extracted using different methods; the method for extracting elements for the current case is determined by first identifying the case type. For example, different cases require different elements to be extracted, and the elements to be extracted for each case can be set according to requirements. In other disclosed embodiments, different case types use the same method for element extraction.

[0027] One approach is to use Named Entity Recognition (NER) to extract elements from the current case. Named entities generally refer to entities in text that have specific meaning or strong referentiality, typically including place names, organization names, dates and times, and proper nouns. Before extracting elements from the descriptive information of the current case, the descriptive information can be preprocessed. NER methods mainly include BILSTM+CRF and BERT+CRF, but are not limited to these two. Preprocessing methods can include word segmentation and stop word removal.

[0028] In some disclosed embodiments, the initial elements, historical elements, and target elements include several entities and at least some of the entity attributes. Optionally, for the current case, the initial elements and target elements contain the same number of entities, and the attributes of each entity may be the same or different. For historical cases, the historical elements and target elements contain the same number of entities, and the attributes of each entity may be the same or different. Optionally, it may also include some relationships between entities. By generating case element relationships, the relationships between cases are increased. Entities and relationships between entities in the case are extracted, and the entities and relationships between entities in different cases are merged to provide a relationship basis for implicit case strings. In some disclosed embodiments, after classifying the current case, element extraction can be performed directly on the preprocessed descriptive information without the need for two preprocessing steps. Optionally, a relationship extraction method (NRE) is used to extract the relationships between entities in the case. NRE (Neural Relation Extraction) refers to extracting relationships between entities from text. NRE methods mainly include BILSTM + ATTENTION, TEXTCNN, and BERT, but are not limited to these three.

[0029] Step S12 above may include the following steps:

[0030] Identify target entities in the current case and several historical cases. A target entity is an entity that exists in both the current case and at least one historical case. In other words, a target entity is an entity that exists in both the current case and some historical cases. The target entity can be identified based on the attributes of each entity. For example, if a vehicle has the same license plate number in two cases, then that vehicle exists in both cases.

[0031] Then, the attributes of the target entity in the current case and / or the target historical cases are merged to update the attributes of the target entity in the current case and / or the target historical cases. The target historical cases are historical cases that include the target entity. In other words, the attributes of the target entity in the target historical cases can be merged with the attributes of the target entity in the current case to update the attributes of the target entity in the current case. Alternatively, the attributes of the target entity in the current case can be merged with the attributes of the target entity in the target historical cases to update the attributes of the target entity in the target historical cases. Or, the attributes of the target entity in the current case and the target historical cases can be mutually merged to update the attributes of the target entity in both the current case and the target historical cases.

[0032] In cases where multiple target entities exist, for each target entity, the step of "merging the attributes of the target entity in the current case and / or target historical cases to update the attributes of the target entity in the current case and / or target historical cases" is performed separately. That is, the attributes of each target entity in the current case and / or target historical cases are updated separately.

[0033] In some disclosed embodiments, the initial element, historical element, and target element also include at least some of the relationships between entities.

[0034] Step S12 may further include the following steps:

[0035] Determine the target relationships in the current case and each historical case. A target relationship is a relationship that does not exist simultaneously in the current case and at least one historical case. Then, based on the target relationships, add new relationships between corresponding entities in the current case or each historical case.

[0036] Specifically, based on the target association, the method of adding the corresponding association between entities in the current case or each historical case includes determining the entity corresponding to the target association, checking whether the entity corresponding to the target association exists in the current case or other historical cases, and if the entity corresponding to the target association exists, establishing the association between the entities as the target association.

[0037] In some disclosed embodiments, the case merging method provided in this disclosure may further include the following steps: constructing an initial case graph based on the initial elements of the current case. Each node in the initial case graph represents an entity in the case. At least some nodes include attributes of the corresponding entity. The initial case graph may be a knowledge graph. A knowledge graph is essentially a semantic network, where nodes represent entities or concepts, and edges represent various semantic relationships between entities / concepts. It can describe cognitive knowledge at various levels, such as concepts, facts, and rules. With its rich semantic representation capabilities and flexible structure, the knowledge graph serves as an effective carrier for representing information and knowledge in the cognitive and physical worlds within the computer world, becoming an important infrastructure for artificial intelligence applications.

[0038] In some disclosed embodiments, the initial case map includes a central node. The central node serves as an identifier for the current case. In other disclosed embodiments, the case identifier may also be a case number or other similar information.

[0039] The method for determining the target entities in the current case and several historical cases can be as follows: Identify target nodes in the historical case graph that represent the same entity as those in the initial case graph. The historical case graph consists of target elements from each historical case, with different nodes corresponding to different entities. The historical case graph includes several historical center nodes. Each center node can be used to represent an identifier for a historical case, and each center node is associated with nodes representing the entity of the corresponding historical case. The historical case graph can be a knowledge graph.

[0040] The above method of determining the target node used to represent the same entity in the historical case map and the initial case map can be based on the attributes of each node. If the attributes of the node are the same in different cases, then the node is determined to be the target node.

[0041] The above-mentioned method of fusing the attributes of the target entity in the current case and / or the target historical cases to update the attributes of the target entity in the current case and / or the target historical cases can be as follows:

[0042] The target nodes in the initial case graph (China) are merged into the corresponding target nodes in the historical case graph, and other nodes in the initial case graph are used as new nodes in the historical case graph. These other nodes are nodes in the historical case graph that do not represent the same entity. In other words, the other nodes are nodes in the initial case graph other than the target nodes.

[0043] Then, the attributes of each target node in the initial case graph are merged with the attributes of the corresponding target nodes in the historical case graph to update the attributes of each target node in the historical case graph. The updated historical case graph includes the initial case graph. In other words, the initial case graph is merged into the historical case graph to update the historical case graph.

[0044] Before performing step S13 above, the following steps may also be performed:

[0045] The updated historical case map is vectorized to obtain the feature vectors for each case. Each case includes the current case and all historical cases. Then, the similarity between cases is determined based on the distance between the feature vectors.

[0046] For example, knowledge graph embedding representation techniques, such as graph convolutional neural networks and graph attention networks, can be used to embed the nodes and relationships in the updated historical case graph to obtain the feature vectors of each case.

[0047] For example, the feature vector of case 1 is N1={x1,x2,x3,x4,x5,x6,…xt}, and the feature vector of case 2 is N2={y1,y2,y3,y4,y5,y6,…yt}. Here, t represents the dimension; for example, if t equals 256, it means that the feature vector of the case has a dimension of 256.

[0048] One way to determine the similarity between cases based on the distance between each feature vector is to use similarity measurement methods, such as cosine angle distance or Euclidean distance, to calculate the distance between each feature vector. The smaller the distance, the more similar the cases are. Following the previous example, the method for calculating the cosine angle distance between N1 and N2 can refer to formula (1):

[0049] (1);

[0050] Here, s1 represents the cosine angle distance between N1 and N2. The cosine angle distance can be directly used as the similarity score, or the similarity score can be transformed to obtain the actual similarity score.

[0051] By constructing a historical case graph and combining it with graph embedding techniques, a vectorized representation of the cases in the graph is obtained, thereby calculating the similarity between cases and completing the construction of the case relationship graph. This method for calculating similar cases does not require setting rules and weights, and is highly flexible.

[0052] In some disclosed embodiments, step S13 above may include the following steps:

[0053] A case relationship graph is determined based on the similarity between cases. Nodes in the case relationship graph represent individual cases. The weights of edges in the graph represent the similarity between the two cases corresponding to that edge. The case relationship graph can be a knowledge graph.

[0054] Then, community detection is performed on the case relationship graph to obtain several candidate communities. Each candidate community includes multiple cases. For example, a community detection algorithm (e.g., the Louvain algorithm) is used to perform community detection on the case relationship graph to obtain potential candidate communities between different cases.

[0055] Next, candidate communities that meet the preset segmentation criteria are designated as target communities. Cases within a target community can be merged. For example, the confidence level of each candidate community is determined. If the confidence level is greater than or equal to a preset confidence level, the candidate community is determined to meet the preset segmentation criteria. For instance, for each candidate community, the density of cases within that community is determined. This density can represent the community confidence level; a higher density indicates a higher confidence level, and vice versa. If the density is less than the preset confidence level, the candidate community is determined not to be a group, and cases within that candidate community cannot be merged. If the density is greater than or equal to the preset confidence level, the candidate community belongs to a group, thus enabling case merging.

[0056] By employing graph algorithm community detection technology, case community groups can be obtained, and relationships between hidden cases with multiple degrees of complexity can be discovered, thus solving the problem of difficulty in linking and merging hidden cases.

[0057] The execution entity of the case merging method can be a case merging device, such as a terminal device, server, or other processing device. The terminal device can be a monitoring device in a security system, a network video recorder, user equipment (UE), mobile device, user terminal, terminal, cellular phone, cordless phone, personal digital assistant (PDA), handheld device, computing device, vehicle-mounted device, wearable device, etc. In some possible implementations, the case merging method can be implemented by a processor calling computer-readable instructions stored in memory.

[0058] Please see Figure 2 , Figure 2 This is a schematic diagram of an embodiment of the case merging device of this application. The case merging device 50 includes an element extraction module 51, an element merging module 52, and a case merging module 53. The element extraction module 51 is used to extract elements from the descriptive information of the current case to obtain several initial elements of the current case; the element merging module 52 is used to merge the initial elements of the current case with historical elements of several historical cases to obtain the target elements of the current case and each historical case; the case merging module 53 is used to merge the current case and several historical cases based on the similarity between the target elements of the current case and the target elements of several historical cases.

[0059] The above-described solution extracts elements from the descriptive information of the current case, then merges the current case and several historical cases based on the similarity between the elements of the current case and the elements of several historical cases. This eliminates the need for manual data analysis, improving the efficiency of case merging. Furthermore, by fusing the initial elements of the current case with the historical elements of several historical cases to obtain the target elements of both the current case and each historical case, and then merging the current case and several historical cases based on the similarity between the target elements of the current case and each historical case, the accuracy of case merging is further enhanced.

[0060] The functions of each module can be found in the case merging method implementation examples, and will not be repeated here.

[0061] Please see Figure 3 , Figure 3 This is a schematic diagram of the structure of an embodiment of the electronic device of this application. The electronic device 60 includes a memory 61 and a processor 62. The processor 62 is used to execute program instructions stored in the memory 61 to implement the steps in any of the above-described case merging method embodiments. In a specific implementation scenario, the electronic device 60 may include, but is not limited to, a microcomputer or a server. In addition, the electronic device 60 may also include mobile devices such as laptops and tablets, which are not limited here.

[0062] Specifically, processor 62 controls itself and memory 61 to implement the steps in any of the above-described case merging method embodiments. Processor 62 may also be referred to as a CPU (Central Processing Unit). Processor 62 may be an integrated circuit chip with signal processing capabilities. Processor 62 may also be a general-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component. A general-purpose processor may be a microprocessor or any conventional processor. Furthermore, processor 62 may be implemented using integrated circuit chips.

[0063] The above-described solution extracts elements from the descriptive information of the current case, then merges the current case and several historical cases based on the similarity between the elements of the current case and the elements of several historical cases. This eliminates the need for manual data analysis, improving the efficiency of case merging. Furthermore, by fusing the initial elements of the current case with the historical elements of several historical cases to obtain the target elements of both the current case and each historical case, and then merging the current case and several historical cases based on the similarity between the target elements of the current case and each historical case, the accuracy of case merging is further enhanced.

[0064] Please see Figure 4 , Figure 4 This is a schematic diagram of a computer-readable storage medium according to an embodiment of the present application. The computer-readable storage medium 70 stores program instructions 71 that can be executed by a processor. The program instructions 71 are used to implement the steps in any of the above-described embodiments of the case merging method.

[0065] The above-described solution extracts elements from the descriptive information of the current case, then merges the current case and several historical cases based on the similarity between the elements of the current case and the elements of several historical cases. This eliminates the need for manual data analysis, improving the efficiency of case merging. Furthermore, by fusing the initial elements of the current case with the historical elements of several historical cases to obtain the target elements of both the current case and each historical case, and then merging the current case and several historical cases based on the similarity between the target elements of the current case and each historical case, the accuracy of case merging is further enhanced.

[0066] In some embodiments, the functions or modules of the apparatus provided in this disclosure can be used to perform the methods described in the above method embodiments. The specific implementation can be referred to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.

[0067] The description of the various embodiments above tends to emphasize the differences between the various embodiments. The similarities or similarities between them can be referred to, and for the sake of brevity, they will not be repeated here.

[0068] In the several embodiments provided in this application, it should be understood that the disclosed methods and apparatus can be implemented in other ways. For example, the apparatus implementations described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms.

[0069] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for merging cases, characterized in that, include: The descriptive information of the current case is extracted to obtain several initial elements of the current case; wherein, the elements include: the object of the behavior, the location of the occurrence, the time of the occurrence, the subject of the behavior, and / or the behavior; The initial elements of the current case are merged with the historical elements of several historical cases to obtain the target elements of the current case and each of the historical cases. Based on the similarity between the target elements of the current case and the target elements of several historical cases, the current case and the several historical cases are merged. The method further includes: Based on the initial elements of the current case, an initial case graph is constructed for the current case. Each node in the case graph represents an entity of the case, and at least some nodes contain the attributes of the corresponding entity. The initial elements of the current case are integrated with the historical elements of several historical cases through the fusion of the initial case map of the current case and the historical case maps of the historical cases.

2. The method according to claim 1, characterized in that, The initial elements, the historical elements, and the target elements include several entities and at least some of the attributes of the entities. The process of fusing the initial elements of the current case with the historical elements of several historical cases to obtain the target elements of the current case and each of the historical cases includes: Identify the target entity in the current case and several historical cases, wherein the target entity is an entity that exists in the current case and at least one historical case; The attributes of the target entity in the current case and / or the target historical case are merged to update the attributes of the target entity in the current case and / or the target historical case, wherein the target historical case is a historical case containing the target entity.

3. The method according to claim 2, characterized in that, The initial elements, the historical elements, and the target elements also include at least some relationships between entities. The step of fusing the initial elements of the current case with the historical elements of several historical cases to obtain the target elements of the current case and each of the historical cases includes: Determine the target association relationships in the current case and each of the historical cases, wherein the target association relationships are association relationships that do not exist simultaneously in the current case and at least one historical case; Based on the target association, add corresponding associations between entities in the current case or each historical case.

4. The method according to claim 2 or 3, characterized in that, The determination of the target entity in the current case and several historical cases includes: Identify target nodes in the historical case map that represent the same entity as those in the initial case map. The historical case map is composed of target elements of each historical case, and different nodes correspond to different entities. The step of fusing the attributes of the target entity in the current case and / or the target historical case to update the attributes of the target entity in the current case and / or the target historical case includes: The target nodes in the initial case graph are merged into the corresponding target nodes in the historical case graph, and other nodes in the initial case graph are used as new nodes in the historical case graph, wherein the other nodes are nodes in the historical case graph that do not represent the same entity. The attributes of each target node in the initial case map are merged with the attributes of the corresponding target nodes in the historical case map to update the attributes of each target node in the historical case map. The updated historical case map includes the initial case map.

5. The method according to claim 4, characterized in that, The initial case map includes a central node, which serves as the identifier for the current case. The historical case map includes several historical central nodes, each of which represents the identifier for a historical case. Each central node is associated with a node representing the entity of the corresponding historical case. The method further includes: The case type is obtained by extracting the type of the current case from the description information of the current case; The identifier for the current case includes the case type of the current case.

6. The method according to claim 4, characterized in that, Before merging the current case and the several historical cases based on the similarity between the target elements of the current case and the target elements of several historical cases, the method further includes: The updated historical case map is vectorized to obtain the feature vector of each case, and each case includes the current case and each of the historical cases. The similarity between the cases is determined based on the distance between each of the feature vectors.

7. The method according to claim 1, characterized in that, The merging of the current case and the historical cases based on the similarity between the target elements of the current case and the target elements of several historical cases includes: Based on the similarity between the cases, a case relationship graph is determined, where nodes in the case relationship graph represent the cases, and the weights of the edges in the relationship graph represent the similarity between the two cases corresponding to the edges. Community detection is performed on the case relationship graph to obtain several candidate communities, each of which includes multiple cases; The candidate communities that meet the preset division conditions are designated as target communities, and cases in the target communities can be merged.

8. The method according to claim 7, characterized in that, Before selecting the candidate communities that meet the preset classification conditions as the target communities, the method further includes: Determine the confidence level of each candidate community; In response to the confidence level being greater than or equal to a preset confidence level, it is determined that the candidate community meets the preset partitioning condition.

9. An electronic device, characterized in that, The method includes a memory and a processor, the processor being configured to execute program instructions stored in the memory to implement the method according to any one of claims 1 to 8.

10. A computer-readable storage medium having program instructions stored thereon, characterized in that, When the program instructions are executed by the processor, they implement the method described in any one of claims 1 to 8.

Images