Event reasoning method, device and computer storage medium

Abstract

The present application provides an event reasoning method, apparatus, and computer storage mechanism, mainly including selecting one or more cognitive nodes in the reasoning network according to the event type of the target event, and determining the target reasoning path of the target event according to the selected one or more cognitive nodes; obtaining the data to be analyzed of the target event according to the attribute information of the target event; performing reasoning analysis on the data to be analyzed by using the target reasoning path to obtain the reasoning result of the target event. Thus, the present application can realize intelligent event clue reasoning analysis according to the event type and attribute information of the target event to improve the safety control efficiency.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of artificial intelligence reasoning technology, and in particular to an event reasoning method, device and computer storage medium. Background Art

[0002] With the further development of the mobile Internet era, contemporary business presents new characteristics in the following aspects. Specifically, in terms of objects, it presents the characteristics of fragmented individual behavior, virtualization and multiplexing of individual identities; in terms of locations, it presents the characteristics of weakening regional attributes; in terms of objects, it presents the characteristics of virtualization; and in terms of organization, it presents the characteristics of unfamiliarity of offline people and looseness of online organizations. Summary of the invention

[0003] In view of the above problems, the present application provides an event reasoning method, device and computer storage medium, which can realize intelligent event clue reasoning analysis to improve event investigation efficiency.

[0004] The first aspect of the present application provides an event reasoning method, which includes: selecting one or more cognitive nodes in an inference network according to the event type of a target event, and determining a target reasoning path of the target event according to the selected one or more cognitive nodes; acquiring data to be analyzed of the target event according to attribute information of the target event; and performing reasoning analysis on the data to be analyzed using the target reasoning path to obtain an inference result of the target event.

[0005] A second aspect of the present application provides a computer storage medium, on which computer instructions are stored. When the computer instructions are executed by a processor, the processor executes the method described in the first aspect.

[0006] The third aspect of the present application provides an event reasoning device, which includes: a matching module, which selects one or more cognitive nodes in the reasoning network according to the event type of the target event, and determines the target reasoning path of the target event according to the selected one or more cognitive nodes; a screening module, which obtains the data to be analyzed of the target event according to the attribute information of the target event; and a reasoning module, which performs reasoning analysis on the data to be analyzed using the target reasoning path to obtain the reasoning result of the target event.

[0007] In summary, the event inference methods, devices, and computer storage media provided by the embodiments of the present application can generate a target inference path for a target event according to the event type of the target event, filter out the data to be analyzed for the target event based on the attribute information of the target event, and perform inference analysis on the data to be analyzed using the target inference path to obtain the inference analysis result of the relevant event clues for the target event. Accordingly, the present application realizes intelligent inference analysis of event clues through technical means such as machine learning and automatic extraction of event elements. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments described in the embodiments of the present application, and those of ordinary skill in the art can also obtain other drawings based on these drawings.

[0009] Figure 1 It is a schematic flowchart of the event inference method according to the first embodiment of the present application.

[0010] Figure 2 It is a schematic flowchart of the event inference method according to the second embodiment of the present application.

[0011] Figure 3 It is a schematic architecture diagram of the event inference device according to the fourth embodiment of the present application.

[0012] Element reference numerals

[0013] 300: Event inference device; 302: Matching module; 304: Screening module; 306: Inference module. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0014] To enable those skilled in the art to better understand the technical solutions in the embodiments of the present application, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art shall fall within the protection scope of the embodiments of the present application.

[0015] First Embodiment

[0016] Figure 1 It shows a schematic flowchart of the event inference method according to the first embodiment of the present application. As shown in the figure, this embodiment mainly includes the following steps:

[0017] Step S102: According to the event type of the target event, select one or more cognitive nodes in the inference network, and determine the target inference path of the target event based on the selected one or more cognitive nodes.

[0018] Optionally, the cognitive nodes in the inference network may include, but are not limited to, an image analysis engine, a trajectory analysis engine, etc.

[0019] In this embodiment, one or more cognitive nodes matching the event type may be selected from the inference network according to the event type of the target event. Wherein, in the case of selecting multiple cognitive nodes, the connection relationship between each cognitive node may be further set to determine the inference logic of the target inference path.

[0020] For example, an image analysis engine and a trajectory analysis engine may be selected to form the target inference path of the target event for analyzing the target object related to the target event and its activity trajectory.

[0021] Optionally, the target features analyzed by each cognitive node may be the same or different.

[0022] Optionally, the target features analyzed by the cognitive node include, but are not limited to, one or more of object features, behavior features, state features, and location features.

[0023] For example, if the cognitive node includes an image analysis engine, the target features analyzed may include object features, behavior features, state features, etc. If the cognitive node includes a trajectory analysis engine, the target features analyzed may include location features, etc.

[0024] Step S104: Obtain the data to be analyzed of the target event according to the attribute information of the target event.

[0025] Optionally, the attribute information of the target event may include at least one of time information and location information.

[0026] In this embodiment, according to the attribute information of the target event, the data collection time (such as the collection period) and the data collection location (such as longitude and latitude) of the data to be analyzed may be determined, and the accessed data attributes may be automatically searched to perform spatial screening and time screening on the data, so as to obtain the data to be analyzed of the target event.

[0027] Optionally, the data to be analyzed may include database data, which includes but is not limited to identity data, vehicle data, asset data, surveillance audio and video data, ticket data, etc.

[0028] Optionally, the data to be analyzed may also include device terminal data, which includes but is not limited to access control data, water, electricity and gas meter data, etc.

[0029] It should be noted that the data to be analyzed is not limited to the above description, and any other data that can be used to assist in performing event reasoning analysis can be applicable to this application.

[0030] For example, assuming that the time information of the data to be analyzed is set to October 1, 2021 to October 7, 2021, and the location information is the Bund, Shanghai, then all monitoring data from October 1, 2021 to October 7, 2021 of various data acquisition devices (such as surveillance cameras) deployed on the Bund, Shanghai can be collected as the data to be analyzed.

[0031] Optionally, the screened data to be analyzed may be standardized, for example, the view data may be structured and clustered, and the unstructured text data may be keyword extracted to form data that can be used by the inference network.

[0032] Step S106, performing reasoning analysis on the data to be analyzed using the target reasoning path to obtain the reasoning result of the target event.

[0033] Optionally, based on each target feature analyzed corresponding to each cognitive node in the target reasoning path, the data to be analyzed can be identified to obtain one or more reasoning results of the target event.

[0034] Specifically, the image analysis engine in the target reasoning path can first be used to identify the object features of the data to be analyzed, so as to infer the target object of the target event. Then, the trajectory analysis engine can be used to identify the location features of the data to be analyzed based on the reasoning results of the image analysis engine, so as to infer the activity trajectory of the target object, and thereby infer event clues such as the location of the target event, organization members, and members' destinations.

[0035] Optionally, the target reasoning path may be used to perform reasoning analysis on the data to be analyzed to obtain one or more reasoning results of the target event.

[0036] Optionally, when multiple inference results of the target event are obtained, each confidence value corresponding to each inference result may also be obtained, and based on each confidence value, each inference result is arranged in sequence, thereby outputting each inference result in sequence.

[0037] For example, when multiple candidate objects of a target event are inferred, each confidence value corresponding to each candidate object may be further output, and each candidate object having a confidence value higher than a preset threshold may be determined as the target object of the target event.

[0038] Optionally, the inference network can adopt a combination of an iterative dilated convolutional neural model and a conditional random field model to obtain feature entities in the data to be analyzed, and adopt a combination of the CloseIE tool and a self-training deep learning model to analyze the feature entities and obtain the inference results of the target event.

[0039] In summary, the event inference method of this embodiment can automatically screen the data to be analyzed according to the event type and attribute information of the target event, and use it to mine the relevant event clues of the target event. Accordingly, the present application can achieve intelligent inference analysis of the target event, which can not only improve the efficiency of event investigation, but also enhance the efficiency of social control and the ability of security prevention.

[0040] Second Embodiment

[0041] Figure 2 The flowchart of the event inference method according to the second embodiment of the present application is shown. As shown in the figure, this embodiment mainly includes the following steps:

[0042] Step S202: According to the type of the target event, sequentially select the location analysis node, object analysis node, and trajectory analysis node in the inference network, and use them to determine the target inference path of the target event.

[0043] Optionally, the event type of the target event can be set. According to the set event type, the location analysis node, object analysis node, and trajectory analysis node can be sequentially selected from the inference network, and according to the selected analysis nodes and the inference logical relationships of the analysis nodes, the target inference path of the inference network can be determined.

[0044] Step S204: According to the time information and location information of the target event, obtain the data to be analyzed of the target event.

[0045] For example, the main crime scene range and main crime time period of the target event can be set, and based on this, the data to be analyzed of the target event can be screened for use in performing the inference analysis of the target event.

[0046] Step S206: Use the location analysis node to infer the group activity area of the target event according to the state characteristics and location characteristics in the data to be analyzed.

[0047] Specifically, the location analysis node can be used to combine the state characteristics and location characteristics in the data to be analyzed for identification, so as to preliminarily determine the activity group of the target event, and then analyze the corresponding activity gathering places of the determined target group, and use this to infer the group activity area of the target event.

[0048] Step S208: Use the object analysis node to identify the object identities in the group activity area according to the object characteristics in the data to be analyzed, so as to infer the target members of the target event.

[0049] Specifically, the object analysis node can be used to analyze each object in the group activity area according to the group activity area inferred by the location analysis node and the object characteristics in the data to be analyzed, so as to dig out the core members of the target event.

[0050] Step S210: Use the trajectory analysis node to analyze the movement trajectories corresponding to each target member according to the object characteristics and position characteristics in the data to be analyzed, so as to infer the landing points corresponding to each target member.

[0051] Specifically, the trajectory analysis node can be used to analyze the movement trajectories and activity rules of each target member according to the target members inferred by the object analysis node and the object characteristics and position characteristics in the data to be analyzed, so as to find the landing points of each target member.

[0052] Step S212: Output each target member and the landing points corresponding to each target member as the inference result of the target event.

[0053] In summary, the event inference method of this embodiment can deeply mine the clues related to the event according to the set event type and event attributes, greatly reduce the time and energy consumed by manual clue research and judgment in the business process, and improve the data investigation efficiency.

[0054] The Third Embodiment

[0055] The third embodiment of the present application provides a computer storage medium, on which computer instructions are stored. When the computer instructions are executed by a processor, the processor executes the method described in the first embodiment or the second embodiment above.

[0056] The Fourth Embodiment

[0057] Figure 3 Fig. shows the schematic architecture of the event inference device according to the third embodiment of the present application. As shown in the figure, the event inference device 300 of this embodiment mainly includes a matching module 302, a screening module 304, and an inference module 306.

[0058] The matching module 302 is configured to select one or more cognitive nodes in the inference network according to the event type of the target event, and determine the target inference path of the target event according to the one or more selected cognitive nodes.

[0059] Optionally, the cognitive nodes at least include an image analysis engine and a trajectory analysis engine.

[0060] Optionally, the target features analyzed by each of the cognitive nodes may be the same or different; wherein, the target features include at least one of object features, behavior features, state features, and position features.

[0061] The data screening module 304 is configured to obtain the data to be analyzed for the target event according to the attribute information of the target event.

[0062] Optionally, the attribute information of the target event includes at least one of time information and location information.

[0063] The inference module 306 is configured to perform inference analysis on the data to be analyzed by using the target inference path, and obtain an inference result of the target event.

[0064] Optionally, the inference module 306 further includes identifying the data to be analyzed based on the target features corresponding to the cognitive nodes in the target inference path, so as to obtain one or more inference results of the target event.

[0065] Optionally, when the inference module 306 obtains multiple inference results of the target event, it further obtains the confidence values corresponding to the inference results; based on the confidence values, the inference results are sorted in sequence, and the inference results are output in order accordingly.

[0066] Optionally, the inference module 306 further includes obtaining feature entities in the data to be analyzed by combining an iterative dilated convolutional neural model and a conditional random field model; analyzing the feature entities by combining a CloseIE tool and a self-training deep learning model to obtain the inference result of the target event.

[0067] Optionally, the inference module 306 further includes sequentially selecting a location analysis node, an object analysis node, and a trajectory analysis node in the inference network according to the type of the target event, so as to determine the target inference path of the target event.

[0068] Optionally, the inference module 306 further includes using the location analysis node to infer the group activity area of the target event according to the state features and location features in the data to be analyzed; using the object analysis node to identify the object identities in the group activity area according to the object features in the data to be analyzed, so as to infer the target members of the target event; using the trajectory analysis node to analyze the movement trajectories corresponding to the target members according to the object features and location features in the data to be analyzed, so as to infer the landing points corresponding to the target members; outputting the target members and the landing points corresponding to the target members as the inference results of the target event.

[0069] In addition, the event inference device 300 according to the embodiment of the present invention can also be used to implement other steps in the foregoing embodiments of the event inference method, and has the beneficial effects of the corresponding method step embodiments, which will not be elaborated herein.

[0070] In summary, the event inference methods, devices, and computer storage media provided by the embodiments of the present application can implement intelligent inference and analysis of target events according to the event types and event attributes of the target events, which can not only greatly reduce the time and energy consumed by manual clue research and judgment in the business process, but also effectively improve the data investigation efficiency to improve the security control efficiency and prevention capabilities.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments 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 recorded in the foregoing embodiments, or perform equivalent replacements for 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.

Claims

1. An event reasoning method, characterized in that, it includes: According to the event type of the target event, select one or more cognitive nodes in the reasoning network, and determine the target reasoning path of the target event according to the selected one or more cognitive nodes; wherein, if multiple cognitive nodes are selected, obtain the connection relationship between the cognitive nodes to determine the reasoning logic of the target reasoning path; According to the attribute information of the target event, obtain the data to be analyzed of the target event; the data to be analyzed includes database data and device terminal data; and Use the target reasoning path to perform reasoning analysis on the data to be analyzed to obtain the reasoning result of the target event; Among them, the obtaining the data to be analyzed of the target event according to the attribute information of the target event includes: according to the attribute information of the target event, determine the data collection time and data collection location of the data to be analyzed, so as to perform spatial screening and time screening on the access data to obtain the data to be analyzed of the target event.

2. The event reasoning method according to claim 1, characterized in that, The cognitive nodes at least include an image analysis engine and a trajectory analysis engine.

3. The event reasoning method according to claim 1, characterized in that, The attribute information of the target event includes at least one of time information and location information.

4. The event reasoning method according to claim 1, characterized in that, The target features corresponding to the analysis of each cognitive node can be the same or different; Among them, the target features include at least one of object features, behavior features, state features, and location features.

5. The event reasoning method according to claim 4, characterized in that, The using the target reasoning path to perform reasoning analysis on the data to be analyzed to obtain the reasoning result of the target event includes: Based on the target features corresponding to the analysis of each cognitive node in the target reasoning path, identify the data to be analyzed to obtain one or more of the reasoning results of the target event.

6. The event reasoning method according to claim 5, characterized in that, The method further includes: When obtaining multiple reasoning results of the target event, also obtain the confidence values corresponding to each reasoning result; Based on each confidence value, arrange each reasoning result in order and output each reasoning result in order accordingly.

7. The event reasoning method according to claim 1, characterized in that, The using the target reasoning path to perform reasoning analysis on the data to be analyzed to obtain the reasoning result of the target event includes: Adopt a combination of an iterative dilation convolutional neural model and a conditional random field model to obtain the feature entities in the data to be analyzed; Adopt a combination of the CloseIE tool and a self-training deep learning model to analyze the feature entities to obtain the reasoning result of the target event.

8. The event reasoning method according to claim 1, characterized in that, Select one or more cognitive nodes in the inference network according to the event type of the target event, and determine the target inference path of the target event according to the selected one or more cognitive nodes. Specifically, it includes: According to the type of the target event, sequentially select a location analysis node, an object analysis node, and a trajectory analysis node in the inference network, and determine the target inference path of the target event accordingly.

9. The event inference method according to claim 8, characterized in that, The performing inference analysis on the data to be analyzed by using the target inference path to obtain an inference result of the target event specifically includes: Using the location analysis node, infer the group activity area of the target event according to the state feature and location feature in the data to be analyzed; Using the object analysis node, identify the object identities in the group activity area according to the object features in the data to be analyzed, so as to infer the target members of the target event; Using the trajectory analysis node, analyze the respective movement trajectories corresponding to the target members according to the object features and the location features in the data to be analyzed, so as to infer the respective landing points corresponding to the target members; Output each of the target members and each of the landing points corresponding to each of the target members as the inference result of the target event.

10. A computer storage medium, characterized in that, Computer instructions are stored on the computer storage medium, and when the computer instructions are executed by a processor, the processor executes the method according to any one of claims 1 to 9.

11. An event inference device, characterized in that, including: A matching module, which selects one or more cognitive nodes in the inference network according to the event type of the target event, and determines the target inference path of the target event according to the selected one or more cognitive nodes; wherein, if multiple cognitive nodes are selected, obtain the connection relationship between the cognitive nodes to determine the inference logic of the target inference path; A screening module, which obtains the data to be analyzed of the target event according to the attribute information of the target event; the data to be analyzed includes database data and device terminal data; An inference module, which performs inference analysis on the data to be analyzed by using the target inference path to obtain an inference result of the target event; wherein, the obtaining the data to be analyzed of the target event according to the attribute information of the target event includes: determining the data collection time and data collection location of the data to be analyzed according to the attribute information of the target event, so as to perform spatial screening and time screening on the access data to obtain the data to be analyzed of the target event.

Images