Event dynamic graph construction method and device

Abstract

The invention provides an event dynamic graph construction method and device. The method comprises the steps of obtaining an event information set of a target event; extracting multiple pieces of entity information from the multiple pieces of event information, constructing multiple entity nodes based on the multiple pieces of entity information, and constructing multiple event nodes based on the multiple pieces of event information; generating an event time sequence chain of all the event nodes based on the event information corresponding to each event node, and determining an event relationship of any two event nodes based on the event time sequence chain and the event information corresponding to each event node; and generating an association edge between each event node and the entity node associated with the event node, and constructing a relationship edge between any two event nodes based on the event relationship between any two event nodes to obtain an event dynamic graph of the target event. By adopting the technical means, the problem that the generated event atlas lacks detailed nodes due to lack of mining of detailed information in the event information in the prior art is solved.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and in particular, to a method and device for constructing an event dynamic graph. Background Art

[0002] Event graphs can represent and understand events and their relationships in text in a structured way, thus supporting various advanced applications and services. For example, hidden patterns and trends can be mined from the graph, such as the development laws of sudden events, the paths of public opinion dissemination, etc. Although there are many current technical challenges, with the continuous progress of natural language processing technology and the increase in computing resources, event graph construction is gradually maturing and showing great application potential in multiple fields. Existing methods for constructing event graphs simply convert event information into graphs, lacking the mining of more detailed information in the event information, resulting in the generated event graphs lacking more detailed nodes. Summary of the Invention

[0003] In view of this, embodiments of this application provide a method, device, electronic device, and computer-readable storage medium for constructing an event dynamic graph to solve the problem that the generated event graphs lack detailed nodes due to the lack of mining of detailed information in event information in the prior art.

[0004] In the first aspect of the embodiments of this application, a method for constructing an event dynamic graph is provided, including: obtaining an event information set of a target event, where the event information set includes multiple pieces of event information about the target event; extracting multiple pieces of entity information from the multiple pieces of event information, constructing multiple entity nodes based on the multiple pieces of entity information, and constructing multiple event nodes based on the multiple pieces of event information, where one entity node corresponds to one piece of entity information and one event node corresponds to one piece of event information; generating an event time sequence chain for all event nodes based on the event information corresponding to each event node, and determining the event relationship between any two event nodes based on the event time sequence chain and the event information corresponding to each event node; generating an association edge between each event node and the entity node associated with the event node, and constructing a relationship edge between any two event nodes based on the event relationship between the any two event nodes to obtain the event dynamic graph of the target event.

[0005] In a second aspect of the embodiments of the present application, an event dynamic graph construction device is provided, including: an acquisition module configured to acquire a set of event information of a target event, where the set of event information includes multiple pieces of event information about the target event; a construction module configured to extract multiple pieces of entity information from the multiple pieces of event information, construct multiple entity nodes based on the multiple pieces of entity information, and construct multiple event nodes based on the multiple pieces of event information, where one entity node corresponds to one piece of entity information and one event node corresponds to one piece of event information; a determination module configured to generate an event time sequence chain of all event nodes based on the event information corresponding to each event node, and determine the event relationship between any two event nodes based on the event time sequence chain and the event information corresponding to each event node; a generation module configured to generate association edges between each event node and the entity nodes associated with the event node, and construct relationship edges between any two event nodes based on the event relationship between any two event nodes, so as to obtain an event dynamic graph of the target event.

[0006] In a third aspect of the embodiments of the present application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the steps of the above method are implemented.

[0007] In a fourth aspect of the embodiments of the present application, a computer-readable storage medium is provided, where the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0008] The beneficial effects of the embodiments of the present application compared with the prior art are as follows: acquiring a set of event information of a target event, where the set of event information includes multiple pieces of event information about the target event; extracting multiple pieces of entity information from the multiple pieces of event information, constructing multiple entity nodes based on the multiple pieces of entity information, and constructing multiple event nodes based on the multiple pieces of event information, where one entity node corresponds to one piece of entity information and one event node corresponds to one piece of event information; generating an event time sequence chain of all event nodes based on the event information corresponding to each event node, and determining the event relationship between any two event nodes based on the event time sequence chain and the event information corresponding to each event node; generating association edges between each event node and the entity nodes associated with the event node, and constructing relationship edges between any two event nodes based on the event relationship between any two event nodes, so as to obtain an event dynamic graph of the target event. By adopting the above technical means, the problem that the generated event graph lacks detailed nodes due to the lack of mining of detailed information in the event information in the prior art can be solved, and further the information expression degree of the event graph can be improved. Description of the Drawings

[0009] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0010] Figure 1 is a schematic flowchart of a method for constructing an event dynamic graph provided by an embodiment of the present application;

[0011] Figure 2 is a schematic flowchart of another method for constructing an event dynamic graph provided by an embodiment of the present application;

[0012] Figure 3 is a schematic structural diagram of an event dynamic graph construction device provided by an embodiment of the present application;

[0013] Figure 4 is a schematic structural diagram of an electronic device provided by an embodiment of the present application. Detailed implementation manners

[0014] In the following description, for the purpose of illustration rather than limitation, specific details such as specific system structures and technologies are presented to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0015] Next, a method and device for constructing an event dynamic graph according to an embodiment of the present application will be described in detail with reference to the accompanying drawings.

[0016] Figure 1 is a schematic flowchart of a method for constructing an event dynamic graph provided by an embodiment of the present application. Figure 1 The method for constructing the event dynamic graph can be executed by a computer or a server, or software on a computer or a server. As Figure 1 shown, the method for constructing the event dynamic graph includes:

[0017] S101, obtaining an event information set of a target event, where the event information set includes multiple pieces of event information about the target event;

[0018] S102, extracting multiple pieces of entity information from the multiple pieces of event information, constructing multiple entity nodes based on the multiple pieces of entity information, and constructing multiple event nodes based on the multiple pieces of event information, where one entity node corresponds to one piece of entity information, and one event node corresponds to one piece of event information;

[0019] S103. Generate the event time sequence chain of all event nodes based on the event information corresponding to each event node, and determine the event relationship between any two event nodes based on the event time sequence chain and the event information corresponding to each event node;

[0020] S104. Generate an association edge between each event node and the entity node associated with the event node, and construct a relationship edge between any two event nodes based on the event relationship between any two event nodes, so as to obtain the event dynamic graph of the target event.

[0021] The event information of the target event can be from news, articles, blogs, and comments, etc. It is all the information generated during the entire process from the occurrence to the end of the target event, and can also include some information related to the target event. For example, if the target event is the release of a certain product, the event information of the target event comes from the preheating before the release of the product, the release, the user feedback on the usage experience, until the popularity of the product is less than the preset popularity and ends.

[0022] Entity information includes time information, location information, and person information, etc. The types of entity information are divided into many kinds, and the common ones include people, institutions, locations, geopolitics, man-made facilities, time, weapons and equipment, positions, means of transportation, etc. A piece of time information can construct a time entity node, a piece of location information can construct a location entity node, and a piece of person information can construct a task entity node. A piece of event information can represent a development stage of the target event, or a small event of the target event, so an event node can be constructed based on a piece of event information.

[0023] Generate the event time sequence chain of all event nodes based on the event information corresponding to each event node, generate an association edge between each event node and the entity node associated with the event node, determine the event relationship between any two event nodes, construct a relationship edge between any two event nodes based on the event relationship between any two event nodes, and finally obtain the event dynamic graph of the target event. The event time sequence chain is a time sequence chain obtained by arranging events in chronological order.

[0024] According to the technical solution provided by the embodiments of the present application, an event information set of a target event is obtained, where the event information set includes multiple pieces of event information about the target event; multiple pieces of entity information are extracted from the multiple pieces of event information, multiple entity nodes are constructed based on the multiple pieces of entity information, and multiple event nodes are constructed based on the multiple pieces of event information, where one entity node corresponds to one piece of entity information and one event node corresponds to one piece of event information; an event time sequence chain of all event nodes is generated based on the event information corresponding to each event node, and the event relationship between any two event nodes is determined based on the event time sequence chain and the event information corresponding to each event node; an association edge is generated between each event node and the entity node associated with the event node, and a relationship edge between the any two event nodes is constructed based on the event relationship between the any two event nodes, so as to obtain an event dynamic graph of the target event. By adopting the above technical means, the problem that the existing technology lacks the mining of detailed information in event information, resulting in the lack of detailed nodes in the generated event graph, can be solved, and further the information expression degree of the event graph can be improved.

[0025] Further, after obtaining the event dynamic graph, the method further includes: obtaining subsequent event information about the target event; constructing new entity nodes and new event nodes based on the subsequent event information; and updating the event dynamic graph by using the new entity nodes and the new event nodes.

[0026] The existing method can only generate an event graph according to the existing information and cannot update the previously generated event graph with the information generated subsequently. The embodiments of the present application can update the event dynamic graph based on the subsequent event information.

[0027] Furthermore, updating the event dynamic graph by using the new entity nodes and the new event nodes includes: determining the event relationship between each new event node and other event nodes based on the event information corresponding to each event node; determining whether there is an association between each new event node and all entity nodes; generating a new association edge between each new event node and the entity node associated with the event node, and constructing a relationship edge between the event node and other event nodes based on the event relationship between each new event node and other event nodes, so as to update the event dynamic graph.

[0028] Determining the event relationship between each new event node and other event nodes, where the other event nodes are any one of the event nodes except the new event node, including the new event node and the previous event nodes. Determining whether there is an association between each new event node and the previous entity nodes and the new entity nodes. For the new event node, a new association edge is generated between the event node and the entity node associated with it, and a relationship edge between the event node and other event nodes is constructed based on the event relationship between the event node and other event nodes, so as to update the event dynamic graph.

[0029] Further, before determining the event relationship between any two event nodes based on the event time sequence chain and the event information corresponding to each event node, the method further includes: using a pre-trained model to judge the authenticity of each piece of event information, and deleting the untrue event information; performing named entity recognition on each piece of event information to obtain the entity information of the entity nodes associated with that piece of event information; using an event extraction algorithm to process each piece of event information to obtain the event core information of that piece of event information; constructing a plurality of entity nodes based on multiple pieces of entity information, and constructing a plurality of event nodes based on multiple pieces of event core information.

[0030] Named entity recognition is to recognize nouns such as people and places in event information. Extracting the event core information of a piece of event information from each piece of event information can be regarded as extracting the summary of that piece of event information, that is, summarizing that piece of event information with short information. The event core information is the content that describes what happened in that piece of event information. Construct an entity node for a piece of entity information and an event node for a piece of event core information.

[0031] Event relationship edge generation. The time sequence relationship can be directly processed using the time element information contained in the event. The causal relationship and the inclusion relationship can be obtained through model classification. For the judgment of the co-reference relationship, rules are needed for assistance. First, it is necessary to judge that the time and place are within a similar range and the event types are the same. Then use the model to judge whether there is co-reference.

[0032] Further, determining the event relationship between any two event nodes based on the event time sequence chain and the event information corresponding to each event node includes: inputting the event time sequence chain and the event information corresponding to any two event nodes into an event relationship diagnosis model to determine the event relationship between the any two event nodes; wherein, the event relationship diagnosis model has been trained to be able to determine the event relationship corresponding to any two pieces of event information based on any two pieces of event information.

[0033] The event relationship diagnosis model can be any commonly used large language model. In order to enable the large language model to determine the event relationship corresponding to any two pieces of event information based on any two pieces of event information, the large language model can be trained as follows:

[0034] Obtain the basic corpus training data, event relationship training data, and event time sequence chain. Among them, the basic corpus training data is the event information of each hot event that has occurred. The event relationship training data is the event information of any two pieces of event information, the labels of any two pieces of event information, and the event time sequence chain of all hot events. The label is the event relationship corresponding to the two pieces of event information, and the event time sequence chain is the time sequence chain of all hot events generated based on the event information of each hot event; Use the basic corpus training data to train the large language model so that the large language model masters the event information of each hot event that has occurred; Use the event relationship training data to train the large language model so that the large language model predicts the event relationship corresponding to the two pieces of event information based on the two pieces of event information and the event time sequence chain, and optimize the large language model based on the prediction result and the label; Obtain the basic corpus data in real time. Among them, the basic corpus data is the event information of each hot event that is currently highly popular and has not ended; Let the large language model master the information in the basic corpus data.

[0035] Further, based on the event relationship between any two event nodes, construct the relationship edge between the two event nodes, including: when the event relationship between any two event nodes is a time sequence relationship, construct a time sequence relationship edge for the two event nodes; when the event relationship between any two event nodes is a causal relationship, construct a causal relationship edge for the two event nodes; when the event relationship between any two event nodes is an inclusion relationship, construct an inclusion relationship edge for the two event nodes; when the event relationship between any two event nodes is a co-reference relationship, merge the two event nodes.

[0036] Event relationships include: time sequence relationship, causal relationship, inclusion relationship, co-reference relationship. Relationship edges include: time sequence relationship edge, causal relationship edge, inclusion relationship edge. The time sequence relationship edge, causal relationship edge, and inclusion relationship edge can be distinguished by different colors. If the time in one piece of event information is earlier than the time in another piece of event information, it can be inferred that the former occurs before the latter, and the event relationship corresponding to the two event nodes is a time sequence relationship. One piece of event information can correspond to a small event. For example, one piece of event information describes that a product has added a new function, and another piece of event information describes that because the product has added a new function, the popularity has increased greatly, then the event relationship corresponding to the two pieces of event information is a causal relationship. For example, one piece of event information describes that a product has added a new function, and another piece of event information describes the application of the new function added to the product, then the event relationship corresponding to the two pieces of event information is an inclusion relationship (the former includes the latter). The co-reference relationship means that the two pieces of event information describe the same event. When the event relationship between any two event nodes is a co-reference relationship, merge the two event nodes.

[0037] Further, multiple entity nodes are constructed based on multiple pieces of entity information, including: for each piece of entity information: when the piece of entity information is time information, a time entity node is constructed; when the piece of entity information is location information, a location entity node is constructed; when the piece of entity information is person information, a person entity node is constructed.

[0038] Entity nodes include: time entity nodes, location entity nodes, and person entity nodes. The time entity nodes, location entity nodes, and person entity nodes can be distinguished by different colors.

[0039] Further, association edges are generated between each event node and the entity nodes associated with the event node, including: for each event node and the entity nodes associated with the event node: when the entity node is a time entity node, a time association edge is generated between the event node and the entity node; when the entity node is a location entity node, a location association edge is generated between the event node and the entity node; when the entity node is a person entity node, a person association edge is generated between the event node and the entity node.

[0040] Association edges also include: time association edges, location association edges, and person association edges. The time association edges, location association edges, and person association edges can be distinguished by different colors.

[0041] Figure 2 It is a schematic flowchart of another method for constructing an event dynamic graph provided by an embodiment of the present application.

[0042] As Figure 2 shown, the method includes:

[0043] S201, obtaining subsequent event information about the target event;

[0044] S202, constructing new entity nodes and new event nodes based on the subsequent event information, and updating the event time sequence chain;

[0045] S203, determining the event relationships between each new event node and other event nodes based on the event information and event time sequence chain corresponding to each event node;

[0046] S204, determining whether there is an association between each new event node and all entity nodes;

[0047] S205, generating new association edges between each new event node and the entity nodes associated with the event node, and constructing relationship edges between the event node and other event nodes based on the event relationships between each new event node and other event nodes, so as to update the event dynamic graph.

[0048] Any combination of the above optional technical solutions can form an optional embodiment of the present application, which will not be elaborated one by one here.

[0049] The following is an apparatus embodiment of the present application, which can be used to execute the embodiments of the present application. For details not disclosed in the apparatus embodiments of the present application, please refer to the embodiments of the present application.

[0050] Figure 3 It is a schematic diagram of an event dynamic graph construction apparatus provided by an embodiment of the present application. As Figure 3 shown, the event dynamic graph construction apparatus includes:

[0051] An acquisition module 301, configured to acquire an event information set of a target event, where the event information set includes multiple pieces of event information about the target event;

[0052] A construction module 302, configured to extract multiple pieces of entity information from the multiple pieces of event information, construct multiple entity nodes based on the multiple pieces of entity information, and construct multiple event nodes based on the multiple pieces of event information, where one entity node corresponds to one piece of entity information, and one event node corresponds to one piece of event information;

[0053] A determination module 303, configured to generate an event time sequence chain of all event nodes based on the event information corresponding to each event node, and determine the event relationship between any two event nodes based on the event time sequence chain and the event information corresponding to each event node;

[0054] A generation module 304, configured to generate an association edge between each event node and an entity node associated with the event node, and construct a relationship edge between the any two event nodes based on the event relationship between the any two event nodes, so as to obtain an event dynamic graph of the target event.

[0055] According to the technical solution provided by the embodiments of the present application, an event information set of a target event is obtained, where the event information set includes multiple pieces of event information about the target event; multiple pieces of entity information are extracted from the multiple pieces of event information, multiple entity nodes are constructed based on the multiple pieces of entity information, and multiple event nodes are constructed based on the multiple pieces of event information, where one entity node corresponds to one piece of entity information and one event node corresponds to one piece of event information; an event time sequence chain of all event nodes is generated based on the event information corresponding to each event node, and the event relationship between any two event nodes is determined based on the event time sequence chain and the event information corresponding to each event node; an association edge is generated between each event node and the entity node associated with the event node, and a relationship edge between the any two event nodes is constructed based on the event relationship between the any two event nodes, so as to obtain an event dynamic graph of the target event. By adopting the above technical means, the problem that the existing technology lacks the mining of detailed information in event information, resulting in the generated event graph lacking detailed nodes, can be solved, and further the information expression degree of the event graph can be improved.

[0056] In some embodiments, the generating module 304 is further configured to obtain subsequent event information about the target event; construct new entity nodes and new event nodes based on the subsequent event information; and update the event dynamic graph by using the new entity nodes and the new event nodes.

[0057] In some embodiments, the generating module 304 is further configured to determine the event relationship between each new event node and other event nodes based on the event information corresponding to each event node; determine whether there is an association between each new event node and all entity nodes; generate a new association edge between each new event node and the entity node associated with the event node, and construct a relationship edge between the event node and other event nodes based on the event relationship between each new event node and other event nodes, so as to update the event dynamic graph.

[0058] In some embodiments, the constructing module 302 is further configured to use a pre-trained model to judge the authenticity of each piece of event information, and delete the untrue event information; perform named entity recognition on each piece of event information to obtain the entity information of the entity node associated with the piece of event information; process each piece of event information by using an event extraction algorithm to obtain the event core information of the piece of event information; construct multiple entity nodes based on the multiple pieces of entity information, and construct multiple event nodes based on the multiple pieces of event core information.

[0059] In some embodiments, the determining module 303 is further configured to input the event time sequence chain and the event information corresponding to any two event nodes into an event relationship diagnosis model to determine the event relationship between the any two event nodes; where the event relationship diagnosis model has been trained and can determine the event relationship corresponding to any two pieces of event information based on any two pieces of event information.

[0060] In some embodiments, the determination module 303 is further configured to obtain basic corpus training data, event relationship training data, and event time sequence chains. The basic corpus training data is the event information of each hot event that has occurred. The event relationship training data is the event information of any two events, the labels of any two pieces of event information, and the event time sequence chains of all hot events. The label is the event relationship corresponding to the two pieces of event information, and the event time sequence chain is the time sequence chain of all hot events generated based on the event information of each hot event. Train the large language model using the basic corpus training data so that the large language model masters the event information of each hot event that has occurred. Train the large language model using the event relationship training data so that the large language model predicts the event relationship corresponding to the two pieces of event information based on the two pieces of event information and the event time sequence chain, and optimize the large language model based on the prediction result and the label. Obtain basic corpus data in real time, where the basic corpus data is the event information of each hot event that is currently highly popular and has not ended. Let the large language model master the information in the basic corpus data.

[0061] In some embodiments, the generation module 304 is further configured to construct a time sequence relationship edge for any two event nodes when the event relationship between them is a time sequence relationship; construct a causal relationship edge for any two event nodes when the event relationship between them is a causal relationship; construct an inclusion relationship edge for any two event nodes when the event relationship between them is an inclusion relationship; and merge the two event nodes when the event relationship between any two event nodes is a co-reference relationship.

[0062] In some embodiments, the construction module 302 is further configured to, for each piece of entity information: construct a time entity node when the piece of entity information is time information; construct a location entity node when the piece of entity information is location information; and construct a person entity node when the piece of entity information is person information.

[0063] In some embodiments, the construction module 302 is further configured to, for each event node and the entity node associated with the event node: generate a time association edge between the event node and the entity node when the entity node is a time entity node; generate a location association edge between the event node and the entity node when the entity node is a location entity node; and generate a person association edge between the event node and the entity node when the entity node is a person entity node.

[0064] The association edges include: time entity nodes, location entity nodes, and person association edges. The time entity nodes, location entity nodes, and person association edges can be distinguished by different colors.

[0065] In some embodiments, the generation module 304 is further configured to obtain subsequent event information about the target event; construct new entity nodes and new event nodes based on the subsequent event information; determine the event relationships between each new event node and other event nodes based on the event information corresponding to each event node; determine whether there is an association between each new event node and all entity nodes; generate new association edges between each new event node and the entity nodes associated with the event node, and construct relationship edges between the event node and other event nodes based on the event relationships between each new event node and other event nodes, so as to update the event dynamic graph.

[0066] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution is prior or subsequent. The execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0067] Figure 4 is a schematic diagram of the electronic device 4 provided by the embodiments of the present application. As Figure 4 shown, the electronic device 4 of this embodiment includes: a processor 401, a memory 402, and a computer program 403 stored in the memory 402 and executable on the processor 401. When the processor 401 executes the computer program 403, the steps in the above various method embodiments are implemented. Alternatively, when the processor 401 executes the computer program 403, the functions of each module / unit in the above various device embodiments are implemented.

[0068] The electronic device 4 may be a desktop computer, a notebook, a palm computer, a cloud server, and other electronic devices. The electronic device 4 may include, but is not limited to, the processor 401 and the memory 402. Those skilled in the art can understand that Figure 4 merely an example of the electronic device 4, does not constitute a limitation to the electronic device 4, and may include more or fewer components than shown in the figure, or different components.

[0069] The processor 401 may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0070] The memory 402 can be an internal storage unit of the electronic device 4. For example, it can be the hard disk or the memory of the electronic device 4. The memory 402 can also be an external storage device of the electronic device 4. For example, it can be a plug-in hard disk equipped on the electronic device 4, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. The memory 402 can also include both the internal storage unit and the external storage device of the electronic device 4. The memory 402 is used to store computer programs and other programs and data required by the electronic device.

[0071] Those skilled in the art can clearly understand that, for the convenience and conciseness of description, only the above division of each functional unit and module is used as an example for illustration. In practical applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0072] If the integrated module / unit is implemented in the form of 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, to implement all or part of the processes in the above embodiment methods of this application, it can also be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, it can implement the steps of the above various method embodiments. The computer program can include computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0073] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for constructing an event dynamic graph, characterized in that: include: Acquire an event information set of a target event, wherein the event information set includes multiple pieces of event information about the target event; Extracting multiple entity information from multiple event information, constructing multiple entity nodes based on the multiple entity information, and constructing multiple event nodes based on the multiple event information, wherein one entity node corresponds to one entity information, and one event node corresponds to one event information; Based on the event information corresponding to each event node, an event sequence chain of all event nodes is generated, and based on the event sequence chain and the event information corresponding to each event node, an event relationship between any two event nodes is determined; An association edge is generated between each event node and an entity node associated with the event node, and based on the event relationship between any two event nodes, a relationship edge between the any two event nodes is constructed to obtain an event dynamic graph of the target event.

2. The method according to claim 1, characterized in that After obtaining the event dynamic graph, the method further includes: Obtaining subsequent event information about the target event; Construct new entity nodes and new event nodes based on subsequent event information; The event dynamic graph is updated with new entity nodes and new event nodes.

3. The method according to claim 2, characterized in that Updating the event dynamic graph using new entity nodes and new event nodes includes: Based on the event information corresponding to each event node, determine the event relationship between each new event node and other event nodes; Determine whether each new event node is associated with all entity nodes; A new association edge is generated between each new event node and an entity node associated with the event node, and based on the event relationship between each new event node and other event nodes, a relationship edge between the event node and other event nodes is constructed to update the event dynamic graph.

4. The method according to claim 1, characterized in that: Before determining the event relationship between any two event nodes based on the event sequence chain and the event information corresponding to each event node, the method further includes: Use the pre-trained model to determine the authenticity of each event information and delete untrue event information; Perform named entity recognition on each piece of event information to obtain entity information of entity nodes associated with the piece of event information; Use the event extraction algorithm to process each event information to obtain the event core information of the event information; Multiple entity nodes are constructed based on multiple entity information, and multiple event nodes are constructed based on multiple event core information.

5. The method according to claim 1, characterized in that Determining the event relationship between any two event nodes based on the event sequence chain and the event information corresponding to each event node includes: Inputting the event sequence chain and event information corresponding to any two event nodes into an event relationship diagnosis model to determine the event relationship between the any two event nodes; The event relationship diagnosis model has been trained and can determine the event relationship corresponding to any two pieces of event information based on the any two pieces of event information.

6. The method according to claim 1, characterized in that Based on the event relationship between any two event nodes, a relationship edge between the any two event nodes is constructed, including: When the event relationship between any two event nodes is a time series relationship, a time series relationship edge is constructed for the any two event nodes; When the event relationship between any two event nodes is a causal relationship, a causal relationship edge is constructed for the any two event nodes; When the event relationship between any two event nodes is an inclusion relationship, an inclusion relationship edge is constructed for the any two event nodes; When the event relationship between any two event nodes is a co-referential relationship, the two event nodes are merged.

7. The method according to claim 1, characterized in that Construct multiple entity nodes based on multiple entity information, including: for each entity information: When the entity information is time information, a time entity node is constructed; When the entity information is location information, a location entity node is constructed; When the entity information is person information, a person entity node is constructed; Generate an associated edge between each event node and an entity node associated with the event node, including: for each event node and an entity node associated with the event node: When the entity node is a time entity node, a time association edge is generated between the event node and the entity node; When the entity node is a location entity node, a location association edge is generated between the event node and the entity node; When the entity node is a person entity node, a person-related edge is generated between the event node and the entity node.

8. An event dynamic graph construction device, characterized in that: include: An acquisition module, configured to acquire an event information set of a target event, wherein the event information set includes a plurality of event information about the target event; A construction module is configured to extract multiple entity information from multiple event information, construct multiple entity nodes based on the multiple entity information, and construct multiple event nodes based on the multiple event information, wherein one entity node corresponds to one entity information and one event node corresponds to one event information; A determination module is configured to generate an event sequence chain of all event nodes based on event information corresponding to each event node, and determine an event relationship between any two event nodes based on the event sequence chain and the event information corresponding to each event node; The generation module is configured to generate an association edge between each event node and an entity node associated with the event node, and based on the event relationship between any two event nodes, construct a relationship edge between the any two event nodes to obtain an event dynamic graph of the target event.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images