Risk early warning method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a risk early warning method and device, electronic equipment and a storage medium, in the embodiment of the invention, the configuration of an identifier of at least one event source channel of a risk scene, a specified time period, a specified geographic area, at least one risk item, at least one target risk identification capability and a push object identifier is supported, so that the risk early warning efficiency is improved. In the risk early warning stage, extracting event information of at least one event occurring in a specified geographic area within a specified time period from event information of events reported by at least one event source channel; and then, through a risk early warning model, performing risk identification on a plurality of risk items by adopting a plurality of risk identification capabilities for event information of events occurring in a specified geographic area in a specified time period of the risk scene, and finally, pushing a risk early warning result of the risk scene to a push object corresponding to the push object identifier. Therefore, multi-dimensional automatic risk early warning is realized, and the risk early warning is more timely and accurate.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular, to a risk warning method, device, electronic device, and storage medium. Background Art

[0002] An event refers to an objective description of a certain behavior or state participated by several roles at a specific time and in a specific environment. Events occur in fields such as urban operation, enterprise operation, and social security. Various roles such as the public, employees, or urban management may report events, and the scale of events is usually relatively large. Among the large-scale events, there may be hot events or sudden events that are prone to risks. It is very necessary to timely discover these events that are prone to risks and conduct risk warnings.

[0003] In traditional risk warning methods, manual methods are mainly used to analyze a large amount of event data, discover events that are prone to risks, and conduct risk warnings. However, this risk warning method is often not timely enough and not accurate enough. Summary of the Invention

[0004] Multiple aspects of this application provide a risk warning method, device, electronic device, and storage medium to make risk warning more timely and accurate.

[0005] An embodiment of this application provides a risk warning method, including: obtaining configuration information of a target risk scenario, where the configuration information includes: identifiers of at least one event source channel, a specified time period, a specified geographical area, at least one risk item, at least one target risk identification ability, and a push object identifier, and different target risk identification abilities are used to identify different risk categories of risk items; obtaining event information of events reported by at least one event source channel according to the identifiers of at least one event source channel, and obtaining event information of at least one target event that occurred in the specified geographical area within the specified time period from the event information of the reported events; inputting the event information of at least one target event into a risk warning model, so as to perform risk identification on at least one risk item by using at least one target risk identification ability through the risk warning model, and obtain a risk warning result of the target risk scenario, where the risk warning result includes the target risk items that appear in the target risk scenario, the event information of the events belonging to the target risk items, and at least one risk category of the target risk items; and pushing the risk warning result of the target risk scenario to the push object corresponding to the push object identifier.

[0006] The embodiment of the present application further provides a risk warning method, which is applied to a cloud server. The method includes: in response to a configuration operation on a configuration page for displaying a target risk scenario in a terminal device, obtaining configuration information of the target risk scenario, where the configuration information includes: identifiers of at least one event source channel, a specified time period, a specified geographical area, at least one risk item, at least one target risk identification capability, and a push object identifier, and different target risk identification capabilities are used to identify different risk categories of risk items; according to the identifiers of at least one event source channel, obtaining event information of events reported by at least one event source channel, and obtaining event information of at least one target event that occurred in the specified geographical area within the specified time period from the event information of the reported events; inputting the event information of at least one target event into a risk warning model, so as to perform risk identification on at least one risk item by using at least one target risk identification capability through the risk warning model, and obtaining a risk warning result of the target risk scenario, where the risk warning result includes the target risk items that appear in the target risk scenario, the event information of the events belonging to the target risk items, and at least one risk category of the target risk items; and pushing the risk warning result of the target risk scenario to the push object corresponding to the push object identifier.

[0007] The embodiment of the present application further provides an electronic device, including: a memory and a processor; the memory is used for storing a computer program; the processor is coupled to the memory and is used for executing the computer program to execute the steps in the risk warning method.

[0008] The embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to be able to implement the steps in the risk warning method.

[0009] The embodiment of the present application provides a risk warning method, device, electronic device and storage medium. In the embodiment of the present application, it is supported to configure the identifiers of at least one event source channel, a specified time period, a specified geographical area, at least one risk item, at least one target risk identification capability and a push object identifier of a risk scenario. In this way, in the risk warning stage, event information of at least one event that occurred in the specified geographical area within the specified time period is extracted from the event information of events reported by at least one event source channel; then, through the risk warning model, the event information of events that occurred in the specified geographical area within the specified time period of the risk scenario is used to perform risk identification on several risk items by using several risk identification capabilities, and finally, the risk warning result of the risk scenario is pushed to the push object corresponding to the push object identifier. Thus, different risk items can be discovered, and the generated risk warning result includes risk types of different dimensions, realizing multi-dimensional automatic risk warning, and the risk warning is more timely and accurate. Description of the Drawings

[0010] The accompanying drawings described herein are used to provide a further understanding of the present application and form a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation to the present application. In the drawings:

[0011] Figure 1 It is a schematic diagram of an application scenario provided by an embodiment of the present application;

[0012] Figure 2 It is a flowchart of a risk warning method provided by an embodiment of the present application;

[0013] Figure 3 It is an exemplary page provided by an embodiment of the present application;

[0014] Figure 4 It is a flowchart of a risk warning method provided by an embodiment of the present application;

[0015] Figure 5 It is a schematic structural diagram of a risk warning device provided by an embodiment of the present application;

[0016] Figure 6 It is a schematic structural diagram of an electronic device provided by an embodiment of the present application. Detailed Embodiments

[0017] To make the objectives, technical solutions, and advantages of the present application clearer, the technical solutions of the present application will be clearly and completely described below in conjunction with the specific embodiments of the present application and the corresponding accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present application.

[0018] In the embodiments of the present application, "at least one" means one or more, and "a plurality" means two or more. "And / or" describes the access relationship of associated objects and indicates that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B may be singular or plural. In the text description of the present application, the character " / " generally represents an "or" relationship between the associated objects before and after. In the embodiments of the present application, "first", "second", "third", etc. are only used to distinguish the content of different objects and have no other special meanings.

[0019] Some terms related to the embodiments of the present application are introduced below:

[0020] Group hot spots: refers to the same appeal that is repeatedly reported multiple times by relevant personnel such as citizens and employees within a certain period of time. That is, the same appeal reported can be understood as a group hot spot. Group hot spots are usually issues that have not been properly handled by the handling department and have been the focus of relevant personnel and have not been resolved for a long time. In this embodiment, the granularity of the time period and the number of repeated feedbacks can be flexibly set. For example, the granularity of the time period is one week, one day, or one hour, and the number of repeated feedbacks is, for example, 100 times.

[0021] Sudden hotspot: refers to the same demand that is reported by citizens, employees and other relevant personnel in multiple different locations within a certain period of time. That is, the same demand reported is usually a sudden hotspot of social concern. In this embodiment, the granularity of the time period, the number of feedback locations, and the number of feedback persons can be flexibly set. For example, the granularity of the time period is one week, one day, or one hour, the number of feedback locations is, for example, 5, and the number of feedback persons is, for example, 100.

[0022] A persistent hotspot refers to a demand that continues to be in a high-incidence state within a certain period of time. A demand that continues to be in a high-incidence state is also a demand with a high feedback frequency. A demand that continues to be in a high-incidence state can be understood as a persistent hotspot. Continuous hotspots can eliminate the influence of time periodic factors to a certain extent, and are a "hard nut to crack" problem at the current stage. In this embodiment, the granularity of the time period and the feedback frequency can be flexibly set. For example, the granularity of the time period is one week, one day, or one hour, and the feedback frequency is, for example, 500 times a day, 500 times an hour, and the like.

[0023] Repeated complaints: refers to the same complaint reported by multiple relevant personnel such as citizens and employees within a certain period of time, that is, the same complaint reported can be understood as a repeated complaint problem. Repeated complaints are usually issues that have not been properly handled by the handling department and are the focus of attention of multiple relevant personnel such as citizens and employees. In this embodiment, the granularity of the time period and the number of people who have reported can be flexibly set. For example, the granularity of the time period is one week, one day, or one hour, and the number of people who have reported is, for example, 100 people.

[0024] High-frequency subject: refers to the subject to which multiple issues are reported. Usually, when multiple relevant personnel such as citizens and employees report multiple issues to the same subject, the subject is the high-frequency subject. The subject can be an event, a complained object, or a topic, but is not limited to this.

[0025] Frequent issues: refers to issues that are repeatedly reported. Usually, there may be frequent issues in the feedback of citizens, employees and other relevant personnel. Monitoring and management of frequent issues in advance can play a "nip in the bud" effect. In this embodiment, the number of repeated feedbacks corresponding to frequent issues can be flexibly set.

[0026] Sensitive appeals: refer to appeals with sensitive negative words that are significantly similar to social negative information. Sensitive appeals may involve issues with a relatively high level of event urgency or significant social impact.

[0027] Concerned topics: usually refer to social issues that are eagerly concerned and urgently need to be solved by relevant personnel such as citizens and employees. In this embodiment, the number of feedback times and the number of people giving feedback on concerned topics can be flexibly set as required.

[0028] Real-time hotspots: refer to events where the number of events within a unit of time exceeds a specified threshold. The unit of time can be, for example, one day or one hour, etc.

[0029] Repeated problems: refer to problems that have occurred and then recurred after an interval of time. Repeated problems may belong to "chronic and difficult" problems. In this embodiment, the number of occurrences of repeated problems and the interval time between two adjacent occurrences can be flexibly set as required.

[0030] Unsolved problems for a long time: refer to problems that have occurred continuously for a period of time and are still occurring recently. Unsolved problems for a long time are usually difficult to completely solve. In this embodiment, the duration of continuous occurrence of unsolved problems for a long time can be flexibly set as required.

[0031] An event refers to an objective description of certain behaviors or states demonstrated by several roles at a specific time and in a specific environment. Some events occur in the fields of urban operation, enterprise operation, social security, etc. Various roles such as the public, employees, or urban management officers may report events, and the scale of events is usually relatively large. Among the large-scale events, there may be hot events or emergency events that are prone to risks. It is very necessary to promptly discover these events prone to risks and conduct risk early warnings. In traditional risk early warning methods, mainly manual methods are used to analyze a large amount of event data to discover events prone to risks and conduct risk early warnings. However, this risk early warning method is often not timely enough and not accurate enough.

[0032] To this end, the embodiments of the present application provide a risk warning method, device, electronic device, and storage medium. In the embodiments of the present application, it is supported to configure the identifiers of at least one event source channel, a specified time period, a specified geographical area, at least one risk item, at least one target risk identification capability, and a push object identifier of a risk scenario. In this way, in the risk warning stage, the event information of at least one event that occurred in the specified geographical area within the specified time period is extracted from the event information of the events reported by at least one event source channel; then, through the risk warning model, the event information of the events that occurred in the specified geographical area during the specified time period of the risk scenario is used to perform risk identification on several risk items by using several risk identification capabilities. Finally, the risk warning result of the risk scenario is pushed to the push object corresponding to the push object identifier. Thus, different risk items can be discovered, and the generated risk warning results include risk types of different dimensions, realizing multi-dimensional automatic risk warning, and the risk warning is more timely and accurate.

[0033] Figure 1 FIG. is a schematic diagram of an application scenario provided by an embodiment of the present application. Refer to Figure 1 , the intelligent research and judgment warning platform located in the cloud is a platform with risk warning functions and can be regarded as a risk warning platform. In the fields of urban operation, enterprise operation, social security, etc., there are often problems such as difficult overall risk perception and difficult discovery of major problems. Difficult overall risk perception can be understood as the difficulty in comprehensively perceiving various events that occur in the fields of urban operation, enterprise operation, social security, etc. With the support of various risk identification capabilities, the intelligent research and judgment warning platform has the following values: real-time discovery of overall risks, accurate research and judgment and timely warning of potential risks and major risks. The intelligent research and judgment warning platform can perform risk discovery, risk perception, and risk warning on urban operation, enterprise operation, social security, specific people, specific places, specific enterprise organizations, etc., but is not limited thereto.

[0034] Specifically, a risk warning model is deployed in the intelligent judgment and warning platform. The risk warning model includes the following multiple sub-models: a group hot spot judgment and warning model with the ability to identify group hot spots, a sudden hot spot judgment and warning model with the ability to identify sudden hot spots, a continuous hot spot judgment and warning model with the ability to identify continuous hot spots, a repeated complaint judgment and warning model with the ability to identify repeated complaints, a high-frequency entity judgment and warning model with the ability to identify high-frequency entities, a hot spot trend judgment and warning model with the ability to identify high-incidence problems, a sensitive appeal judgment and warning model with the ability to identify sensitive appeals, a concerned topic judgment and warning model with the ability to identify concerned topics, a real-time hot spot judgment and warning model with the ability to identify real-time hot spots, a recurrent problem judgment and warning model with the ability to identify recurrent problems, and a long-pending problem judgment and warning model with the ability to identify long-pending problems. These sub-models in the risk warning model are algorithm models with corresponding risk identification capabilities after model training, and these sub-models will be introduced in the following content.

[0035] In practical applications, the intelligent judgment and warning platform supports users to flexibly configure risk scenarios, which are the scenarios that require risk warnings. For example, scenarios such as urban appearance and environment, social security, road problems, and consumption disputes. For example, the user triggers the configuration page of the risk scenario displayed on the terminal device. In the configuration page of the risk scenario, the user can configure configuration information such as the identifier of the event source channel of the risk scenario, the specified time period, the specified geographical area, the required risk identification capabilities, the risk matters that require risk warnings, and the identifier of the push object. The terminal device obtains the configuration information of the risk scenario in response to the user's configuration operation on the configuration page of the risk scenario. Refer to Figure 1 As shown in ① of Figure 1 As shown in ② of , the intelligent judgment and warning platform conducts risk warnings based on the configuration information of the risk scenario. Specifically, the intelligent judgment and warning platform obtains event information reported by various event source channels such as official websites, social network platforms, intelligent environmental sanitation systems, integrated management systems, or citizen service systems, and obtains the event information of several events that occurred in the specified geographical area within the specified time period from the reported event information. It uses the sub-model corresponding to the configured risk identification capabilities in the risk warning model to conduct risk identification, and obtains the risk warning result of the risk scenario. The risk warning result of the risk scenario includes the risk matters that appear in the risk scenario, the event information of the events belonging to the risk matters, and at least one risk category of the risk matters, etc. Refer to Figure 1As shown in ③ in [reference], the intelligent judgment and early warning platform can also generate an early warning push record based on the early warning result of the risk scenario and push it to the push object corresponding to the push object identifier. The push object can be various service systems. For example, service system 1 is a smart environmental sanitation system, an integrated management system, or a citizen service system; service system 2 is a construction vehicle system, a digital city management system, or an integrated command system.

[0036] It should be noted that Figure 1 The application scenarios shown are only exemplary application scenarios, and the embodiments of the present application do not limit the application scenarios. The embodiments of the present application do not Figure 1 limit the devices included in [reference], nor do they Figure 1 limit the positional relationship between the devices in [reference].

[0037] In addition, the terminal device can be hardware or software. When the terminal device is hardware, the user device is, for example, a mobile phone, a tablet computer, a desktop computer, a wearable intelligent device, a smart home device, etc. When the terminal device is software, it can be installed in the above-listed hardware devices. At this time, the terminal device is, for example, multiple software modules or a single software module, etc., and the embodiments of the present application do not limit this.

[0038] The following will detail the technical solutions provided by the embodiments of the present application with reference to the accompanying drawings.

[0039] Figure 2 This is a flowchart of a risk early warning method provided by an embodiment of the present application. Refer to Figure 2 The method may include the following steps:

[0040] 201. Obtain the configuration information of the target risk scenario, where the configuration information includes: the identifier of at least one event source channel, a specified time period, a specified geographical area, at least one target risk identification ability, at least one risk item, and a push object identifier. Different target risk identification abilities are used to identify different risk categories of risk items.

[0041] 202. According to the identifier of at least one event source channel, obtain the event information of the events reported by at least one event source channel, and obtain the event information of at least one target event that occurred in the specified geographical area within the specified time period from the event information of the reported events.

[0042] 203. Input the event information of at least one target event into the risk early warning model to perform risk identification on at least one risk item by using at least one target risk identification ability through the risk early warning model, and obtain the risk early warning result of the target risk scenario, where the risk early warning result includes the target risk item that appears in the target risk scenario, the event information of the event belonging to the target risk item, and at least one risk category of the target risk item.

[0043] 204. Push the risk warning result of the target risk scenario to the push target corresponding to the push target identifier.

[0044] In this embodiment, the target risk scenario refers to the scenario that needs to be subject to risk warning, such as but not limited to: urban appearance scenario, social security scenario, road problem scenario, consumption dispute scenario, etc. In practical applications, users can flexibly configure the target risk scenario as needed. The configuration information of the target risk scenario includes, for example, but not limited to: the identifier of at least one event source channel, a specified time period, a specified geographical area, at least one target risk identification ability, at least one risk matter, and a push target identifier.

[0045] In practical applications, the identifier of the event source channel required for the target risk scenario can be flexibly configured as needed. The event source channel refers to the channel that generates event information. The event source channel includes, for example, but not limited to: official websites, social network platforms, intelligent environmental sanitation systems, integrated law enforcement systems, or citizen service systems, etc.

[0046] In this embodiment, the specified time period is used to define the time period for which risk warning needs to be carried out, and the specified geographical area is used to define the geographical area for which risk warning needs to be carried out. That is, when carrying out risk warning for the target risk scenario, the event information of several events that occurred in the specified geographical area within the specified time period is extracted from the event information of the events reported by the event source channel, and the target risk scenario is subject to risk warning based on the event information of the extracted events.

[0047] In this embodiment, the risk matter refers to the matter that may have risks, and users can configure the risk matters for which the target risk scenario needs to carry out risk warning as needed. The risk matters include, for example, but not limited to: random dumping, sewage manhole covers, medical assistance for disadvantaged groups, communication equipment maintenance problems, drainage facility blockages, overhead cable problems.

[0048] In this embodiment, at least one target risk identification capability is configured for the target risk scenario, and the at least one target risk identification capability is selected from multiple risk identification capabilities possessed by the risk early warning model. It can be understood that the risk identification capability can also be understood as the risk identification function of the risk early warning model. For example, the multiple risk identification capabilities possessed by the risk early warning model are respectively: group hot spot identification capability (which can also be called group hot spot identification function), sudden hot spot identification capability (which can also be called sudden hot spot identification function), continuous hot spot identification capability (which can also be called continuous hot spot identification function), repeated complaint identification capability (which can also be called repeated complaint identification function), high-frequency subject identification capability (which can also be called high-frequency subject identification function), high-incidence problem identification capability (which can also be called high-incidence problem identification function), sensitive appeal identification capability (which can also be called sensitive appeal identification function), concerned topic identification capability (which can also be called concerned topic identification function), real-time hot spot identification capability (which can also be called real-time hot spot identification function), recurrent problem identification capability (which can also be called recurrent problem identification function), long-pending problem identification capability (which can also be called long-pending problem identification function), etc. The target risk identification capability is one or more of the above 11 capabilities, and there is no limitation on this.

[0049] It should be noted that different risk identification capabilities are used to identify different risk categories of risk events. For example, the risk categories are respectively: group hot spots that can be identified by the group hot spot identification capability, sudden hot spots that can be identified by the sudden hot spot identification capability, continuous hot spots that can be identified by the continuous hot spot identification capability, repeated complaint problems that can be identified by the repeated complaint identification capability, high-frequency subjects that can be identified by the high-frequency subject identification capability, high-incidence problems that can be identified by the high-incidence problem identification capability, sensitive appeals that can be identified by the sensitive appeal identification capability, concerned topics that can be identified by the concerned topic identification capability, real-time hot spots that can be identified by the real-time hot spot identification capability, recurrent problems that can be identified by the recurrent problem identification capability, and long-pending problems that can be identified by the long-pending problem identification capability.

[0050] In this embodiment, a push object identifier is configured for the target risk scenario. The push object corresponding to the push object identifier has a need to obtain the risk early warning result of the target risk scenario, and the push object performs corresponding disposal operations based on the risk early warning result of the target risk scenario.

[0051] In some alternative embodiments, to facilitate the configuration of risk scenarios and improve the reliability of risk scenario configuration, a configuration page for risk scenarios is provided for users to complete the configuration operations of risk scenarios. Based on this, for the configuration process of the target risk scenario, the configuration page of the target risk scenario is displayed. The configuration page includes a scenario definition configuration item, a policy configuration item, a data source configuration item, and an information push configuration item; in response to the configuration operation for the scenario definition configuration item, the scenario identifier, at least one target risk identification capability, and at least one risk matter of the target risk scenario are configured, where at least one target risk identification capability is selected from multiple risk identification capabilities possessed by the risk early warning model; in response to the configuration operation for the policy configuration item, the specified time period and the specified geographical area of the target risk scenario are configured; in response to the configuration operation for the data source configuration item, at least one event source channel identifier of the target risk scenario is configured; in response to the configuration operation for the information push configuration item, the push object identifier corresponding to the target risk scenario is configured.

[0052] Take Figure 3 the configuration page of the urban appearance shown as an example. Through the scenario definition configuration item, the risk scenario definition can be completed. Using the scenario definition configuration item of the urban appearance, the risk scenario identifier of the urban appearance can be configured as a1b2c3, and the risk identification capabilities of the urban appearance can be configured as group hotspot identification capability, sudden hotspot identification capability, high-frequency subject identification capability, sensitive appeal identification capability, etc. Figure 3 The risk identification capabilities checked in are the risk identification capabilities required for the urban appearance. The risk matters of the urban appearance include, for example, but are not limited to: random dumping, sewage manhole covers, medical assistance for disadvantaged groups, communication equipment maintenance problems, drainage facility blockages, and overhead cable problems.

[0053] In addition, when the user operates on the policy configuration item in the configuration page of the urban appearance, the specified time period and the specified geographical area of the urban appearance can be configured. When the user operates on the data source configuration item in the configuration page of the urban appearance, at least one event source channel identifier of the urban appearance can be configured; when the user operates on the information push configuration item in the configuration page of the urban appearance, the push object identifier of the urban appearance can be configured.

[0054] In practical applications, when the user has a need to configure the target risk scenario, the user can trigger the terminal device to display the configuration page of the target risk scenario. In some alternative embodiments, to meet the user's configuration requirements for different risk scenarios, the implementation method of displaying the configuration page of the target risk scenario is: display the risk scenario management page, and the risk scenario management page includes page areas of multiple risk scenarios; in response to the trigger operation on the scenario configuration control in the page area of the target risk scenario, display the configuration page of the target risk scenario.

[0055] Take Figure 3Taking the risk scenario management page shown as an example, the risk scenario management page includes a page area for the city appearance and environment, a page area for social security, and so on. When the user clicks on the scenario configuration control in the page area for the city appearance and environment, a configuration page for the city appearance and environment will pop up for the user to configure the city appearance and environment. When the user clicks on the scenario configuration control in the page area for social security, a configuration page for social security will pop up for the user to configure social security.

[0056] Further optionally, in order to meet the user's needs for viewing and modifying the configuration information of the risk scenario, after the target risk scenario is configured, in response to a trigger operation on the scenario details control in the page area of the target risk scenario, a scenario details page of the target risk scenario is displayed; in response to a modification operation triggered in the scenario details page, the configuration information of the target risk scenario in the scenario details page is modified to obtain the modified configuration information of the target risk scenario. For example, when the user clicks on the scenario details control in the page area for the city appearance and environment, a scenario details page for the city appearance and environment will pop up, and the user can modify the configuration information of the city appearance and environment in the scenario details page.

[0057] In practical applications, after the configuration of the target risk scenario is completed, the scenario identifier and configuration information of the target risk scenario are associated and stored, so that the stored configuration information can be obtained based on the scenario identifier of the target risk scenario.

[0058] In this embodiment, after obtaining the configuration information of the target risk scenario, the identifiers of at least one event source channel, a specified time period, and a specified geographical area are obtained from the configuration information, and based on the identifiers of at least one event source channel, the event information of the events reported by at least one event source channel is obtained. Then, the event information of the reported events is filtered according to the specified time period and the specified geographical area, that is, the event information of at least one target event that occurred within the specified time period in the specified geographical area is obtained from the event information of the reported events.

[0059] In this embodiment, at least one target risk identification capability and at least one risk matter of the target risk scenario are obtained from the configuration information of the target risk scenario. After the event information of at least one target event is input into the risk warning model, the risk warning model uses at least one target risk identification capability to perform risk identification on at least one risk matter to obtain the risk warning result of the target risk scenario, where the risk warning result includes the target risk matter that appears in the target risk scenario, the event information of the event belonging to the target risk matter, and at least one risk category of the target risk matter.

[0060] Specifically, the risk warning model can identify whether the configured risk matters occur in the target risk scenario. Herein, the risk matters that occur in the identified target risk scenario are referred to as target risk matters. For example, in the urban appearance and environment scenario, multiple risk matters are configured, such as random dumping, sewage manhole covers, medical assistance for disadvantaged groups, communication equipment maintenance problems, drainage facility blockages, and overhead cable problems. The identified target risk matters are random dumping, sewage manhole covers, medical assistance for disadvantaged groups, and overhead cable problems, etc.

[0061] In this embodiment, the risk warning model also uses at least one target risk identification ability to identify the risk categories of the target risk matters. In practical applications, if a risk matter does not belong to the risk category that can be identified by the target risk identification ability, the risk category of this risk matter is classified as a general matter. When the risk category of a risk matter is a general matter, the probability of the risk caused by the risk matter is relatively small. The at least one target risk identification ability is, for example, group hot spot identification ability, sudden hot spot identification ability, and sensitive appeal identification ability. The risk warning model identifies the risk category of random dumping as a group hot spot, the risk warning model identifies the sewage manhole cover as a sudden hot spot, and the risk warning model identifies the medical assistance for disadvantaged groups as a sensitive appeal; the risk warning model identifies the high-voltage cable problem as a general matter.

[0062] In this embodiment, the risk warning model can also identify the events belonging to the target risk matters, as well as the event information of the events belonging to the target risk matters. The event information of an event includes, for example, but is not limited to: the event occurrence time, the event occurrence location, the object being complained about, and the event content, etc.

[0063] Further optionally, for better risk warning, the risk warning model includes multiple sub-models with different risk identification abilities. Different sub-models have different risk identification abilities, and each sub-model has one risk identification ability. The sub-models are obtained by training the model with the corresponding risk identification ability as the goal, and the trained sub-models use the corresponding risk identification ability to identify risks. The model structure of the sub-models is not limited, for example, including but not limited to: Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), and Long Short-Term Memory (LSTM).

[0064] In the training phase of the sub-model, training data is prepared. The training data includes event information of at least one event in the sample risk scenario and annotation results. The annotation results include risk matters that appear in the sample risk scenario, event information of events belonging to the risk matters that appear, and at least one risk category of the risk matters that appear. The event information of at least one event in the sample risk scenario is input into the sub-model to obtain the prediction results output by the sub-model. Based on the annotation results and prediction results of the sample risk scenario, the loss value is calculated, and the model parameters of the sub-model are adjusted with the goal of minimizing the loss value. The model training is iteratively executed multiple times until the training termination condition is met. The training termination condition can be that the number of model training times reaches the specified number, or the loss value has reached the minimum value. For more introduction about model training, reference can be made to the related technologies.

[0065] In this embodiment, if the risk warning model includes multiple sub-models with different risk identification capabilities, the event information of at least one target event is input into the risk warning model, so that the risk warning model uses at least one target risk identification capability to identify risks for at least one risk matter, and the implementation manner of obtaining the risk warning result of the target risk scenario is as follows: The event information of at least one target event is input into the risk warning model, so that the sub-models corresponding to at least one target risk identification capability respectively perform risk identification; according to the risk identification results respectively output by the sub-models corresponding to at least one target risk identification capability, the risk warning result of the target risk scenario is generated.

[0066] Specifically, the sub-model can identify the target risk matters that appear in the target risk scenario, the event information of events belonging to the target risk matters, and at least one risk category of the target risk matters. The risk identification results respectively output by each sub-model are summarized to obtain the risk warning result of the target risk scenario.

[0067] Several optional sub-models are introduced below.

[0068] 1. Group hot spot research and judgment warning model: It has the ability to identify group hot spots and is used to identify whether the risk category of a risk matter is a group hot spot.

[0069] When building a model for group hot spot research and judgment warning, first, prepare the training data. The training data includes the event information and annotation results of at least one event in the sample risk scenario. The annotation results include the risk matters that appear in the sample risk scenario, the event information of the events belonging to the risk matters that appear, and whether the risk category of the risk matters that appear belongs to the group hot spot. Specifically, the event information of the event includes, for example, but is not limited to: the event occurrence time, the event occurrence location, the object being complained about, and the event content, etc. The risk matters include, for example, but are not limited to: random dumping, sewage manhole covers, medical assistance for people in difficulty, communication equipment maintenance problems, drainage facility blockages, road waterlogging, drainage facility blockages, overhead cable problems. For example, the risk category of the risk matter of random dumping belongs to the group hot spot; the risk categories of risk matters such as sewage manhole covers and medical assistance for people in difficulty do not belong to the group hot spot.

[0070] Next, use the training data to train the group hot spot research and judgment warning model. During the model training process, input the event information of at least one event in the sample risk scenario into the group hot spot research and judgment warning model to obtain the prediction results output by the group hot spot research and judgment warning model. Calculate the loss value based on the annotation results and prediction results of the sample risk scenario, and adjust the model parameters of the group hot spot research and judgment warning model with the goal of minimizing the loss value. Iteratively execute the model training multiple times until the training end condition is met. The training end condition can be that the number of model training times reaches the specified number, or the loss value has reached the minimum value. For more introduction to model training, reference can be made to related technologies.

[0071] When calculating the loss value based on the annotation results and prediction results of the sample risk scenario, use the annotation results as the input to the loss function and the prediction results as the predicted values input to the loss function. The loss function calculates the loss value based on the input true values and predicted values. For example, the risk matters that appear in the sample risk scenario in the annotation results, the event information of the events belonging to the risk matters that appear, and whether the risk category of the risk matters that appear belongs to the group hot spot, etc. are used as the true values respectively, and the risk matters that appear in the sample risk scenario in the prediction results, the event information of the events belonging to the risk matters that appear, and whether the risk category of the risk matters that appear belongs to the group hot spot, etc. are used as the predicted values corresponding to the true values respectively. For example, the risk matters that appear in the sample risk scenario in the annotation results, whether the risk category of the risk matters that appear belongs to the group hot spot, etc. are used as the true values respectively, and the risk matters that appear in the sample risk scenario in the prediction results, whether the risk category of the risk matters that appear belongs to the group hot spot, etc. are used as the predicted values corresponding to the true values respectively.

[0072] In this embodiment, any loss function can be used to determine the loss value of the training data. Any loss function, for example, includes but is not limited to: mean squared error loss function, L2 loss function (also known as Euclidean distance), L1 loss function (Manhattan distance), cross-entropy loss function, softmax loss function, and Focal loss function mainly for solving the problem of imbalance between easy and hard samples, etc.

[0073] 2. Sudden hotspot research and judgment early warning model: It has the ability to identify sudden hotspots and is used to identify whether the risk category of a risk event is a sudden hotspot.

[0074] When dealing with the sudden hotspot research and judgment early warning model, first, prepare the training data. The training data includes the event information and annotation results of at least one event in the sample risk scenario. The annotation results include the risk events that appear in the sample risk scenario, the event information of the events belonging to the risk events that appear, and whether the risk category of the risk events that appear belongs to a sudden hotspot. Specifically, the event information of an event, for example, includes but is not limited to: event occurrence time, event occurrence location, the object being complained about, and event content, etc. Risk events, for example, include but are not limited to: random dumping, sewage manhole covers, medical assistance for people in difficulty, communication equipment maintenance problems, drainage facility blockages, road waterlogging, drainage facility blockages, overhead cable problems. For example, the risk category of the risk event of sewage manhole covers belongs to a sudden hotspot; the risk categories of risk events such as random dumping and medical assistance for people in difficulty do not belong to sudden hotspots.

[0075] Next, use the training data to train the sudden hotspot research and judgment early warning model. During the model training process, input the event information of at least one event in the sample risk scenario into the sudden hotspot research and judgment early warning model to obtain the prediction results output by the sudden hotspot research and judgment early warning model. Calculate the loss value based on the annotation results and prediction results of the sample risk scenario, and take minimizing the loss value as the goal to adjust the model parameters of the sudden hotspot research and judgment early warning model. Iteratively execute the model training multiple times until the end training condition is met. The end training condition can be that the number of model training times reaches the specified number, or the loss value has reached the minimum. For more introductions about model training, reference can be made to related technologies.

[0076] When calculating the loss value based on the annotation result and the prediction result of the sample risk scenario, the annotation result is used as the input to the loss function, and the prediction result is used as the predicted value input to the loss function. The loss function calculates the loss value based on the input true value and predicted value. For example, the risk events that appear in the sample risk scenario in the annotation result, the event information of the events that belong to the risk events that appear, and whether the risk category of the risk events that appear belongs to a sudden hot spot, etc. are used as the true values respectively, and the risk events that appear in the sample risk scenario in the prediction result, the event information of the events that belong to the risk events that appear, and whether the risk category of the risk events that appear belongs to a sudden hot spot, etc. are used as the predicted values corresponding to the true values respectively. For example, the risk events that appear in the sample risk scenario in the annotation result, whether the risk category of the risk events that appear belongs to a sudden hot spot, etc. are used as the true values respectively, and the risk events that appear in the sample risk scenario in the prediction result, whether the risk category of the risk events that appear belongs to a sudden hot spot, etc. are used as the predicted values corresponding to the true values respectively.

[0077] In this embodiment, any loss function can be used to determine the loss value of the training data. Any loss function, for example, includes but is not limited to: mean square error loss function, L2 loss function (also known as Euclidean distance), L1 loss function (Manhattan distance), cross-entropy loss function, softmax loss function, and Focal loss function mainly for solving the problem of imbalance between easy and hard samples, etc.

[0078] 3. Continuous hot spot research and judgment warning model: It has the ability to identify continuous hot spots and is used to identify whether the risk category of a risk event is a continuous hot spot.

[0079] When dealing with the continuous hot spot research and judgment warning model, first, prepare the training data. The training data includes the event information of at least one event in the sample risk scenario and the annotation result. The annotation result includes the risk events that appear in the sample risk scenario, the event information of the events that belong to the risk events that appear, and whether the risk category of the risk events that appear belongs to a continuous hot spot. Specifically, the event information of the event, for example, includes but is not limited to: event occurrence time, event occurrence location, the object being complained about, and event content, etc. Risk events, for example, include but are not limited to: random dumping, sewage manhole covers, medical assistance for disadvantaged groups, communication equipment maintenance problems, drainage facility blockages, road waterlogging, drainage facility blockages, and overhead cable problems. For example, the risk category of the risk event of sewage manhole covers belongs to a continuous hot spot; the risk categories of risk events such as random dumping and medical assistance for disadvantaged groups do not belong to continuous hot spots.

[0080] Next, use the training data to train the continuous hot spot research and judgment early warning model. During the model training process, input the event information of at least one event in the sample risk scenario into the continuous hot spot research and judgment early warning model to obtain the prediction results output by the continuous hot spot research and judgment early warning model. Calculate the loss value based on the annotation results and prediction results of the sample risk scenario, and adjust the model parameters of the continuous hot spot research and judgment early warning model with the goal of minimizing the loss value. Iteratively execute the model training multiple times until the end training condition is met. The end training condition can be that the number of model training times reaches the specified number, or the loss value has reached the minimum value. For more introductions about model training, reference can be made to related technologies.

[0081] When calculating the loss value based on the annotation results and prediction results of the sample risk scenario, use the annotation results as the input to the loss function and the prediction results as the predicted values input to the loss function. The loss function calculates the loss value based on the input true values and predicted values. For example, the risk events that appear in the sample risk scenario in the annotation results, the event information of the events belonging to the risk events that appear, and whether the risk category of the risk events that appear belongs to the continuous hot spot, etc. are used as the true values respectively, and the risk events that appear in the sample risk scenario in the prediction results, the event information of the events belonging to the risk events that appear, and whether the risk category of the risk events that appear belongs to the continuous hot spot, etc. are used as the predicted values corresponding to the true values respectively. For example, the risk events that appear in the sample risk scenario in the annotation results, whether the risk category of the risk events that appear belongs to the continuous hot spot, etc. are used as the true values respectively, and the risk events that appear in the sample risk scenario in the prediction results, whether the risk category of the risk events that appear belongs to the continuous hot spot, etc. are used as the predicted values corresponding to the true values respectively.

[0082] In this embodiment, any loss function can be used to determine the loss value of the training data. Any loss function, for example, includes but is not limited to: mean square error loss function, L2 loss function (also known as Euclidean distance), L1 loss function (Manhattan distance), cross-entropy loss function, softmax loss function, and Focal loss function mainly used to solve the problem of imbalance between easy and difficult samples, etc.

[0083] 4. Repeated complaint research and judgment early warning model: It has the ability to identify repeated complaints and is used to identify whether the risk category of the risk event is a repeated complaint problem.

[0084] When analyzing and predicting the repeated complaint warning model, first, prepare the training data. The training data includes the event information and annotation results of at least one event in the sample risk scenario. The annotation results include the risk matters that occur in the sample risk scenario, the event information of the events belonging to the risk matters that occur, and whether the risk category of the risk matters that occur belongs to repeated complaints. Specifically, the event information of the event includes, for example, but is not limited to: the event occurrence time, the event occurrence location, the object of complaint, and the event content, etc. The risk matters include, for example, but are not limited to: random dumping, sewage manhole covers, medical assistance for disadvantaged groups, communication equipment maintenance problems, drainage facility blockages, road waterlogging, drainage facility blockages, and overhead cable problems. For example, the risk category of the risk matter of drainage facility blockage belongs to repeated complaints; the risk categories of risk matters such as random dumping and medical assistance for disadvantaged groups do not belong to repeated complaints.

[0085] Next, use the training data to train the repeated complaint warning model. During the model training process, input the event information of at least one event in the sample risk scenario into the repeated complaint warning model to obtain the prediction results output by the repeated complaint warning model. Calculate the loss value based on the annotation results and prediction results of the sample risk scenario, and adjust the model parameters of the repeated complaint warning model with the goal of minimizing the loss value. Iteratively execute the model training multiple times until the training end condition is met. The training end condition can be that the number of model training times reaches the specified number, or the loss value has reached the minimum value. For more introduction to model training, reference can be made to related technologies.

[0086] When calculating the loss value based on the annotation results and prediction results of the sample risk scenario, use the annotation results as the input to the loss function, and use the prediction results as the predicted values input to the loss function. The loss function calculates the loss value based on the input true values and predicted values. For example, the risk matters that occur in the sample risk scenario in the annotation results, the event information of the events belonging to the risk matters that occur, and whether the risk category of the risk matters that occur belongs to repeated complaints, etc. are used as the true values respectively, and the risk matters that occur in the sample risk scenario in the prediction results, the event information of the events belonging to the risk matters that occur, and whether the risk category of the risk matters that occur belongs to repeated complaints, etc. are used as the predicted values corresponding to the true values respectively. For example, the risk matters that occur in the sample risk scenario in the annotation results, and whether the risk category of the risk matters that occur belongs to repeated complaints, etc. are used as the true values respectively, and the risk matters that occur in the sample risk scenario in the prediction results, and whether the risk category of the risk matters that occur belongs to repeated complaints, etc. are used as the predicted values corresponding to the true values respectively.

[0087] In this embodiment, any loss function can be used to determine the loss value of the training data. Any loss function, for example, includes but is not limited to: mean squared error loss function, L2 loss function (also known as Euclidean distance), L1 loss function (Manhattan distance), cross-entropy loss function, softmax loss function, and Focal loss function mainly for solving the problem of imbalance between easy and hard samples, etc.

[0088] 5. High-frequency entity judgment and early warning model: It has the ability to identify high-frequency entities and is used to identify whether the risk category of a risk event is a high-frequency entity.

[0089] When judging and warning the high-frequency entity judgment model, first, prepare the training data. The training data includes the event information and annotation results of at least one event in the sample risk scenario. The annotation results include the risk events that occur in the sample risk scenario, the event information of the events belonging to the occurred risk events, and whether the risk category of the occurred risk events belongs to high-frequency entities. Specifically, the event information of an event, for example, includes but is not limited to: event occurrence time, event occurrence location, the object being complained about, and event content, etc. Risk events, for example, include but are not limited to: random dumping, sewage manhole covers, medical assistance for disadvantaged groups, communication equipment maintenance problems, drainage facility blockages, road waterlogging, drainage facility blockages, and overhead cable problems. For example, the risk category of the risk event of communication equipment maintenance problems belongs to high-frequency entities; the risk categories of risk events such as random dumping and medical assistance for disadvantaged groups do not belong to high-frequency entities.

[0090] Next, use the training data to train the high-frequency entity judgment and early warning model. During the model training process, input the event information of at least one event in the sample risk scenario into the high-frequency entity judgment and early warning model to obtain the prediction results output by the high-frequency entity judgment and early warning model. Calculate the loss value based on the annotation results and prediction results of the sample risk scenario, and adjust the model parameters of the high-frequency entity judgment and early warning model with the goal of minimizing the loss value. Iteratively execute the model training multiple times until the end training condition is met. The end training condition can be that the number of model training times reaches the specified number, or the loss value has reached the minimum. For more introductions about model training, reference can be made to related technologies.

[0091] When calculating the loss value based on the annotation result and the prediction result of the sample risk scenario, the annotation result is used as the input to the loss function, and the prediction result is used as the predicted value input to the loss function. The loss function calculates the loss value based on the input true value and predicted value. For example, the risk events that appear in the sample risk scenario in the annotation result, the event information of the events that belong to the risk events that appear, and whether the risk category of the risk events that appear belongs to a high-frequency entity, etc. are used as the true values respectively, and the risk events that appear in the sample risk scenario in the prediction result, the event information of the events that belong to the risk events that appear, and whether the risk category of the risk events that appear belongs to a high-frequency entity, etc. are used as the predicted values corresponding to the true values respectively. For example, the risk events that appear in the sample risk scenario in the annotation result, whether the risk category of the risk events that appear belongs to a high-frequency entity, etc. are used as the true values respectively, and the risk events that appear in the sample risk scenario in the prediction result, whether the risk category of the risk events that appear belongs to a high-frequency entity, etc. are used as the predicted values corresponding to the true values respectively.

[0092] In this embodiment, any loss function can be used to determine the loss value of the training data. Any loss function, for example, includes but is not limited to: mean square error loss function, L2 loss function (also known as Euclidean distance), L1 loss function (Manhattan distance), cross-entropy loss function, softmax loss function, and Focal loss function mainly for solving the problem of imbalance between easy and difficult samples, etc.

[0093] 6. Hot trend research and judgment early warning model: It has the ability to identify high-incidence problems and is used to identify whether the risk category of a risk event is a high-incidence problem.

[0094] When training the hot trend research and judgment early warning model, first, prepare the training data. The training data includes the event information of at least one event in the sample risk scenario and the annotation result. The annotation result includes the risk events that appear in the sample risk scenario, the event information of the events that belong to the risk events that appear, and whether the risk category of the risk events that appear belongs to a high-incidence problem. Specifically, the event information of the event, for example, includes but is not limited to: event occurrence time, event occurrence location, the object being complained about, and event content, etc. Risk events, for example, include but are not limited to: random dumping, sewage manhole covers, medical assistance for disadvantaged groups, communication equipment maintenance problems, drainage facility blockages, overhead cable problems, drainage facility blockages, overhead cable problems. For example, the risk category of the risk event of communication equipment maintenance problems belongs to a high-incidence problem; the risk categories of risk events such as random dumping and medical assistance for disadvantaged groups do not belong to high-incidence problems.

[0095] Next, use the training data to train the hotspot trend research and judgment early warning model. During the model training process, input the event information of at least one event in the sample risk scenario into the hotspot trend research and judgment early warning model to obtain the prediction result output by the hotspot trend research and judgment early warning model. Calculate the loss value based on the annotation result and the prediction result of the sample risk scenario, and take minimizing the loss value as the goal to adjust the model parameters of the hotspot trend research and judgment early warning model. Iteratively execute the model training multiple times until the end training condition is met. The end training condition can be that the number of model training times reaches the specified number, or the loss value has reached the minimum value. For more introductions about model training, reference can be made to related technologies.

[0096] When calculating the loss value based on the annotation result and the prediction result of the sample risk scenario, use the annotation result as the input to the loss function, and use the prediction result as the predicted value to input into the loss function. The loss function calculates the loss value based on the input true value and predicted value. For example, the risk matters that appear in the sample risk scenario in the annotation result, the event information of the events belonging to the risk matters that appear, and whether the risk category of the risk matters that appear belongs to a high-incidence problem, etc. are respectively used as the true values, and the risk matters that appear in the sample risk scenario in the prediction result, the event information of the events belonging to the risk matters that appear, and whether the risk category of the risk matters that appear belongs to a high-incidence problem, etc. are respectively used as the predicted values corresponding to the true values. For example, the risk matters that appear in the sample risk scenario in the annotation result, whether the risk category of the risk matters that appear belongs to a high-incidence problem, etc. are respectively used as the true values, and the risk matters that appear in the sample risk scenario in the prediction result, whether the risk category of the risk matters that appear belongs to a high-incidence problem, etc. are respectively used as the predicted values corresponding to the true values.

[0097] In this embodiment, any loss function can be used to determine the loss value of the training data. Any loss function, for example, includes but is not limited to: mean square error loss function, L2 loss function (also known as Euclidean distance), L1 loss function (Manhattan distance), cross-entropy loss function, softmax loss function, and Focal loss function mainly used to solve the problem of imbalance between easy and difficult samples, etc.

[0098] 7. Sensitive appeal research and judgment early warning model: It has the ability to identify sensitive appeals and is used to identify whether the risk category of a risk matter is a sensitive appeal.

[0099] When evaluating and warning the sensitive appeal judgment model, first, prepare the training data. The training data includes the event information and annotation results of at least one event in the sample risk scenario. The annotation results include the risk matters that appear in the sample risk scenario, the event information of the events belonging to the risk matters that appear, and whether the risk category of the risk matters that appear belongs to the sensitive appeal. Specifically, the event information of the event includes, for example, but is not limited to: the event occurrence time, the event occurrence location, the object being complained about, and the event content, etc. The risk matters include, for example, but are not limited to: random dumping, sewage manhole covers, medical assistance for needy people, communication equipment maintenance problems, drainage facility blockages, road waterlogging, drainage facility blockages, and overhead cable problems. For example, the risk category of the risk matter of medical assistance for needy people belongs to the sensitive appeal; the risk categories of risk matters such as random dumping and sewage manhole covers do not belong to the sensitive appeal.

[0100] Next, use the training data to train the sensitive appeal judgment and warning model. During the model training process, input the event information of at least one event in the sample risk scenario into the sensitive appeal judgment and warning model, and obtain the prediction result output by the sensitive appeal judgment and warning model. Calculate the loss value based on the annotation result and the prediction result of the sample risk scenario, and take minimizing the loss value as the goal to adjust the model parameters of the sensitive appeal judgment and warning model. Iteratively execute the model training multiple times until the end training condition is met. The end training condition can be that the number of model training times reaches the specified number, or the loss value has reached the minimum value. For more introductions about model training, reference can be made to the related technologies.

[0101] When calculating the loss value based on the annotation result and the prediction result of the sample risk scenario, use the annotation result as the input to the loss function, and use the prediction result as the predicted value to input into the loss function. The loss function calculates the loss value based on the input true value and predicted value. For example, the risk matters that appear in the sample risk scenario in the annotation result, the event information of the events belonging to the risk matters that appear, and whether the risk category of the risk matters that appear belongs to the sensitive appeal, etc. are respectively used as the true values, and the risk matters that appear in the sample risk scenario in the prediction result, the event information of the events belonging to the risk matters that appear, and whether the risk category of the risk matters that appear belongs to the sensitive appeal, etc. are respectively used as the predicted values corresponding to the true values. For example, the risk matters that appear in the sample risk scenario in the annotation result, whether the risk category of the risk matters that appear belongs to the sensitive appeal, etc. are respectively used as the true values, and the risk matters that appear in the sample risk scenario in the prediction result, whether the risk category of the risk matters that appear belongs to the sensitive appeal, etc. are respectively used as the predicted values corresponding to the true values.

[0102] In this embodiment, any loss function can be used to determine the loss value of the training data. Any loss function, for example, includes but is not limited to: mean square error loss function, L2 loss function (also known as Euclidean distance), L1 loss function (Manhattan distance), cross-entropy loss function, softmax loss function, and Focal loss function mainly used to solve the problem of imbalance between easy and hard samples, etc.

[0103] 8. Concerned topic research and judgment early warning model: It has the ability to identify concerned topics and is used to identify whether the risk category of a risk event is a concerned topic.

[0104] When training the concerned topic research and judgment early warning model, first, prepare the training data. The training data includes the event information and annotation results of at least one event in the sample risk scenario. The annotation results include the risk events that appear in the sample risk scenario, the event information of the events belonging to the risk events that appear, and whether the risk category of the risk events that appear belongs to the concerned topic. Specifically, the event information of the event, for example, includes but is not limited to: event occurrence time, event occurrence location, the object being complained about, and event content, etc. Risk events, for example, include but are not limited to: random dumping, sewage manhole covers, medical assistance for needy people, communication equipment maintenance problems, drainage facility blockages, road waterlogging, drainage facility blockages, overhead cable problems. For example, the risk category of the risk event of communication equipment maintenance problems belongs to the concerned topic; the risk categories of risk events such as sewage manhole covers and medical assistance for needy people do not belong to the concerned topic.

[0105] Next, use the training data to train the concerned topic research and judgment early warning model. During the model training process, input the event information of at least one event in the sample risk scenario into the concerned topic research and judgment early warning model to obtain the prediction results output by the concerned topic research and judgment early warning model. Calculate the loss value based on the annotation results and prediction results of the sample risk scenario, and take minimizing the loss value as the goal to adjust the model parameters of the concerned topic research and judgment early warning model. Iteratively execute the model training multiple times until the end training condition is met. The end training condition can be that the number of model training times reaches the specified number, or the loss value has reached the minimum. For more introduction to model training, reference can be made to related technologies.

[0106] When calculating the loss value based on the annotation result and the prediction result of the sample risk scenario, the annotation result is used as the input to the loss function, and the prediction result is used as the predicted value input to the loss function. The loss function calculates the loss value based on the input true value and predicted value. For example, the risk events that appear in the sample risk scenario in the annotation result, the event information of the events belonging to the risk events that appear, and whether the risk category of the risk events that appear belongs to a concerned topic, etc. are used as the true values respectively, and the risk events that appear in the sample risk scenario in the prediction result, the event information of the events belonging to the risk events that appear, and whether the risk category of the risk events that appear belongs to a concerned topic, etc. are used as the predicted values corresponding to the true values respectively. For example, the risk events that appear in the sample risk scenario in the annotation result, whether the risk category of the risk events that appear belongs to a concerned topic, etc. are used as the true values respectively, and the risk events that appear in the sample risk scenario in the prediction result, whether the risk category of the risk events that appear belongs to a concerned topic, etc. are used as the predicted values corresponding to the true values respectively.

[0107] In this embodiment, any loss function can be used to determine the loss value of the training data. Any loss function, for example, includes but is not limited to: mean square error loss function, L2 loss function (also known as Euclidean distance), L1 loss function (Manhattan distance), cross-entropy loss function, softmax loss function, and Focal loss function mainly for solving the problem of imbalance between easy and hard samples, etc.

[0108] 9. Real-time hot topic research and judgment early warning model: It has the ability to identify real-time hot topics and is used to identify whether the risk category of a risk event is a real-time hot topic.

[0109] When building the real-time hot topic research and judgment early warning model, first, prepare the training data. The training data includes the event information of at least one event in the sample risk scenario and the annotation result. The annotation result includes the risk events that appear in the sample risk scenario, the event information of the events belonging to the risk events that appear, and whether the risk category of the risk events that appear belongs to a real-time hot topic. Specifically, the event information of the event, for example, includes but is not limited to: event occurrence time, event occurrence location, the object being complained about, and event content, etc. Risk events, for example, include but are not limited to: random dumping, sewage manhole covers, medical assistance for people in difficulty, communication equipment maintenance problems, drainage facility blockages, road waterlogging, drainage facility blockages, overhead cable problems. For example, the risk category of the risk event of drainage facility blockage belongs to a real-time hot topic; the risk categories of risk events such as sewage manhole covers and medical assistance for people in difficulty do not belong to real-time hot topics.

[0110] Next, use the training data to train the real-time hot-spot research and judgment warning model. During the model training process, input the event information of at least one event in the sample risk scenario into the real-time hot-spot research and judgment warning model to obtain the prediction result output by the real-time hot-spot research and judgment warning model. Calculate the loss value based on the annotation result and the prediction result of the sample risk scenario, and take minimizing the loss value as the goal to adjust the model parameters of the real-time hot-spot research and judgment warning model. Iteratively execute the model training multiple times until the end training condition is met. The end training condition can be that the number of model training times reaches the specified number, or the loss value has reached the minimum value. For more introduction about model training, reference can be made to related technologies.

[0111] When calculating the loss value based on the annotation result and the prediction result of the sample risk scenario, use the annotation result as the input to the loss function and the prediction result as the predicted value input to the loss function. The loss function calculates the loss value based on the input true value and predicted value. For example, the risk matters that appear in the sample risk scenario in the annotation result, the event information of the events belonging to the risk matters that appear, and whether the risk category of the risk matters that appear belongs to the real-time hot spot, etc. are respectively used as the true values, and the risk matters that appear in the sample risk scenario in the prediction result, the event information of the events belonging to the risk matters that appear, and whether the risk category of the risk matters that appear belongs to the real-time hot spot, etc. are respectively used as the predicted values corresponding to the true values. For example, the risk matters that appear in the sample risk scenario in the annotation result, whether the risk category of the risk matters that appear belongs to the real-time hot spot, etc. are respectively used as the true values, and the risk matters that appear in the sample risk scenario in the prediction result, whether the risk category of the risk matters that appear belongs to the real-time hot spot, etc. are respectively used as the predicted values corresponding to the true values.

[0112] In this embodiment, any loss function can be used to determine the loss value of the training data. Any loss function, for example, includes but is not limited to: mean squared error loss function, L2 loss function (also known as Euclidean distance), L1 loss function (Manhattan distance), cross-entropy loss function, softmax loss function, and Focal loss function mainly used to solve the problem of imbalance between easy and difficult samples, etc.

[0113] 10. Repeated occurrence problem research and judgment warning model: It has the ability to identify repeated occurrence problems and is used to identify whether the risk category of a risk matter is a repeated occurrence problem.

[0114] When building a model for the recurrent problem judgment and early warning model, first, prepare the training data. The training data includes the event information and annotation results of at least one event in the sample risk scenario. The annotation results include the risk matters that appear in the sample risk scenario, the event information of the events belonging to the risk matters that appear, and whether the risk category of the risk matters that appear belongs to the recurrent problem. Specifically, the event information of the event includes, for example, but is not limited to: the event occurrence time, the event occurrence location, the object being complained about, and the event content, etc. The risk matters include, for example, but are not limited to: random dumping, sewage manhole covers, medical assistance for needy people, communication equipment maintenance problems, drainage facility blockages, road waterlogging, drainage facility blockages, and overhead cable problems. For example, the risk category of the risk matter of random dumping belongs to the recurrent problem; the risk categories of risk matters such as sewage manhole covers and medical assistance for needy people do not belong to the recurrent problem.

[0115] Next, use the training data to train the recurrent problem judgment and early warning model. During the model training process, input the event information of at least one event in the sample risk scenario into the recurrent problem judgment and early warning model, and obtain the prediction results output by the recurrent problem judgment and early warning model. Calculate the loss value based on the annotation results and prediction results of the sample risk scenario, and adjust the model parameters of the recurrent problem judgment and early warning model with the goal of minimizing the loss value. Iteratively execute the model training multiple times until the training end condition is met. The training end condition can be that the number of model training times reaches the specified number, or the loss value has reached the minimum value. For more introductions about model training, reference can be made to the related technologies.

[0116] When calculating the loss value based on the annotation results and prediction results of the sample risk scenario, use the annotation results as the input to the loss function, and use the prediction results as the predicted values to input into the loss function. The loss function calculates the loss value based on the input true values and predicted values. For example, the risk matters that appear in the sample risk scenario in the annotation results, the event information of the events belonging to the risk matters that appear, and whether the risk category of the risk matters that appear belongs to the recurrent problem, etc. are used as the true values respectively, and the risk matters that appear in the sample risk scenario in the prediction results, the event information of the events belonging to the risk matters that appear, and whether the risk category of the risk matters that appear belongs to the recurrent problem, etc. are used as the predicted values corresponding to the true values respectively. For example, the risk matters that appear in the sample risk scenario in the annotation results, and whether the risk category of the risk matters that appear belongs to the recurrent problem, etc. are used as the true values respectively, and the risk matters that appear in the sample risk scenario in the prediction results, and whether the risk category of the risk matters that appear belongs to the recurrent problem, etc. are used as the predicted values corresponding to the true values respectively.

[0117] In this embodiment, any loss function can be used to determine the loss value of the training data. Any loss function, for example, includes but is not limited to: mean squared error loss function, L2 loss function (also known as Euclidean distance), L1 loss function (Manhattan distance), cross-entropy loss function, softmax loss function, and Focal loss function mainly used to solve the problem of imbalance between easy and hard samples, etc.

[0118] 11. Long-pending issue research and judgment early warning model: It has the ability to identify long-pending issues and is used to identify whether the risk category of a risk event is a long-pending issue.

[0119] When training the long-pending issue research and judgment early warning model, first, prepare the training data. The training data includes the event information and annotation results of at least one event in the sample risk scenario. The annotation results include the risk events that appear in the sample risk scenario, the event information of the events belonging to the risk events that appear, and whether the risk category of the risk events that appear belongs to a long-pending issue. Specifically, the event information of an event, for example, includes but is not limited to: event occurrence time, event occurrence location, the object being complained about, and event content, etc. Risk events, for example, include but are not limited to: random dumping, sewage manhole covers, medical assistance for needy people, communication equipment maintenance problems, drainage facility blockages, road waterlogging, drainage facility blockages, overhead cable problems. For example, the risk category of the risk event of road waterlogging belongs to a long-pending issue; the risk categories of risk events such as sewage manhole covers and medical assistance for needy people do not belong to long-pending issues.

[0120] Next, use the training data to train the long-pending issue research and judgment early warning model. During the model training process, input the event information of at least one event in the sample risk scenario into the long-pending issue research and judgment early warning model to obtain the prediction results output by the long-pending issue research and judgment early warning model. Calculate the loss value based on the annotation results and prediction results of the sample risk scenario, and take minimizing the loss value as the goal to adjust the model parameters of the long-pending issue research and judgment early warning model. Iteratively execute the model training multiple times until the end training condition is met. The end training condition can be that the number of model training times reaches the specified number, or the loss value has reached the minimum. For more introduction to model training, reference can be made to related technologies.

[0121] When calculating the loss value based on the annotation result and the prediction result of the sample risk scenario, the annotation result is used as the input to the loss function, and the prediction result is used as the predicted value input to the loss function. The loss function calculates the loss value based on the input true value and predicted value. For example, the risk events that appear in the sample risk scenario in the annotation result, the event information of the events that belong to the risk events that appear, and whether the risk category of the risk events that appear belongs to long-pending issues, etc. are used as the true values respectively, and the risk events that appear in the sample risk scenario in the prediction result, the event information of the events that belong to the risk events that appear, and whether the risk category of the risk events that appear belongs to long-pending issues, etc. are used as the predicted values corresponding to the true values respectively. For example, the risk events that appear in the sample risk scenario in the annotation result, and whether the risk category of the risk events that appear belongs to long-pending issues, etc. are used as the true values respectively, and the risk events that appear in the sample risk scenario in the prediction result, and whether the risk category of the risk events that appear belongs to long-pending issues, etc. are used as the predicted values corresponding to the true values respectively.

[0122] In this embodiment, any loss function can be used to determine the loss value of the training data. Any loss function, for example, includes but is not limited to: mean square error loss function, L2 loss function (also known as Euclidean distance), L1 loss function (Manhattan distance), cross-entropy loss function, softmax loss function, and Focal loss function mainly for solving the problem of imbalance between easy and difficult samples, etc.

[0123] In the embodiment of the present application, the push object identifier is obtained from the configuration information of the target risk scenario, and after obtaining the risk warning result of the target risk scenario output by the risk warning model, the risk warning result of the target risk scenario is pushed to the push object corresponding to the push object identifier.

[0124] Further optionally, more information about the target risk event can be analyzed to guide the push object to better handle the risk. Based on this, the implementation method of pushing the risk warning result of the target risk scenario to the push object corresponding to the push object identifier is: for the target risk event that appears in the target risk scenario, perform clustering analysis on the event information of the events that belong to the target risk event, and generate the warning push content of the target risk scenario according to the clustering analysis result; generate the warning push record of the target risk event according to the event name of the target risk scenario, the number of target events that belong to the risk event, at least one risk category, and the warning push content; push the warning push records of each target risk event in the target risk scenario to the push object corresponding to the push object identifier.

[0125] In practical applications, clustering analysis can be performed on the event information of events belonging to target risk matters. Clustering analysis can be based on the event occurrence time of events belonging to target risk matters. According to the clustering analysis results, determine the number of events belonging to target risk matters that occur at each event occurrence time. Summarize the event content of the events corresponding to the top N event occurrence times with the largest number of events. The obtained content is the early warning push content belonging to target risk matters, where N is an integer set as needed. See Figure 3 , for cases of random dumping, the early warning push content is: Complaint about sewage discharge at the main entrance at 9 o'clock; Sewer discharge problem of a certain barbecue restaurant at 10 o'clock; Illegal sewage discharge in a certain community at 11 o'clock.

[0126] In practical applications, clustering analysis can also be performed on the event information of events belonging to target risk matters based on the event occurrence location of events belonging to target risk matters. According to the clustering analysis results, determine the number of events belonging to target risk matters that occur at each event occurrence location. Summarize the event content of the events corresponding to the top N event occurrence locations with the largest number of events. The obtained content is the early warning push content belonging to target risk matters. See Figure 3 , for cases of sewage manhole covers, the early warning push content is: Problem of a noisy manhole cover in the sewer at the entrance; Manhole cover rupture near a certain intersection.

[0127] In practical applications, clustering analysis can also be performed on the event information of events belonging to target risk matters based on the complained-against object of events belonging to target risk matters. According to the clustering analysis results, determine the number of events belonging to target risk matters that occur for each complained-against object. Summarize the event content of the events corresponding to the top N complained-against objects with the largest number of events. The obtained content is the early warning push content belonging to target risk matters. See Figure 3 , for cases of medical assistance for disadvantaged groups, the early warning push content is: Emergency problems in the high-tech zone; Problems in the rescue station; Inconvenient living in the dormitory. For cases of high-voltage cables, the early warning push content is: Electricity meter replacement problem, Cable drop at the entrance, High-voltage cable disturbing residents.

[0128] In this embodiment, after obtaining the early warning push content of each target risk item that appears in the target risk scenario, an early warning push record of the target risk item can be generated according to the event name of the target risk scenario, the number of target events belonging to the risk item, at least one risk category, and the early warning push content; and the early warning push records of each target risk item in the target risk scenario are pushed to the push object corresponding to the push object identifier. Optionally, a task ID (Identity Document, unique code) can also be assigned to the early warning push record, and the scenario identifier and task ID of the target risk scenario are added to the early warning push record, and the early warning push record is added to the early warning message table. Taking the task ID as the query granularity, query the early warning push records that have not been pushed in the early warning message table, push the early warning push records that have not been pushed to the push object corresponding to the push object identifier, and update the corresponding early warning push records in the early warning message table to the pushed state.

[0129] In some alternative embodiments, to facilitate viewing of the early warning push records, the page area of the target risk scenario may further include an early warning push record control. In response to a trigger operation on the early warning push record control in the page area of the target risk scenario, an early warning push record page of the target risk scenario is displayed, and the early warning push record page includes the early warning push records of the target risk items. Figure 3 For example, when the user clicks on the early warning push record control in the page area of the city appearance, an early warning push record page of the city appearance pops up, and this page displays the early warning push records of each risk item. The early warning push record of each risk item includes information such as the push time, the item name of the risk item, the risk category, the number of events, and the early warning push content.

[0130] The risk early warning method provided by the embodiments of the present application supports configuring the identifiers of at least one event source channel of the risk scenario, a specified time period, a specified geographical area, at least one risk item, at least one target risk identification capability, and a push object identifier. In this way, in the risk early warning stage, the event information of at least one event that occurred within the specified time period in the specified geographical area is extracted from the event information of the events reported by at least one event source channel; then, through the risk early warning model, the event information of the events that occurred in the specified geographical area within the specified time period of the risk scenario is used to perform risk identification for several risk items using several risk identification capabilities. Finally, the risk early warning result of the risk scenario is pushed to the push object corresponding to the push object identifier. Thus, different risk items can be discovered, and the risk early warning results generated include risk types of different dimensions, realizing multi-dimensional automatic risk early warning, and the risk early warning is more timely and accurate.

[0131] Figure 4 It is a flowchart of a risk early warning method provided by the embodiments of the present application. Applied to a cloud server, seeFigure 4 , the method may include the following steps:

[0132] 401. In response to a configuration operation on a configuration page for displaying a target risk scenario in a terminal device, obtain configuration information of the target risk scenario, where the configuration information includes: identifiers of at least one event source channel, a specified time period, a specified geographical area, at least one risk item, at least one target risk identification capability, and a push object identifier, and different target risk identification capabilities are used to identify different risk categories of risk items.

[0133] 402. According to the identifiers of at least one event source channel, obtain event information of events reported by at least one event source channel, and obtain event information of at least one target event that occurred in the specified geographical area within the specified time period from the event information of the reported events.

[0134] 403. Input the event information of at least one target event into a risk warning model, so as to perform risk identification on at least one risk item by using at least one target risk identification capability through the risk warning model, and obtain a risk warning result of the target risk scenario, where the risk warning result includes the target risk items that appear in the target risk scenario, the event information of the events belonging to the target risk items, and at least one risk category of the target risk items.

[0135] 404. Push the risk warning result of the target risk scenario to the push object corresponding to the push object identifier.

[0136] For the detailed implementation processes and technical effects of the steps in the embodiments of the present application, reference may be made to the relevant descriptions in the foregoing embodiments, which will not be elaborated herein.

[0137] Figure 5 It is a schematic structural diagram of a risk warning device provided by an embodiment of the present application. Refer to Figure 5 , the device may include:

[0138] A first acquisition module 51, configured to acquire configuration information of a target risk scenario, where the configuration information includes: identifiers of at least one event source channel, a specified time period, a specified geographical area, at least one risk item, at least one target risk identification capability, and a push object identifier, and different target risk identification capabilities are used to identify different risk categories of risk items;

[0139] A second acquisition module 52, configured to acquire event information of events reported by at least one event source channel according to the identifiers of at least one event source channel, and acquire event information of at least one target event that occurred in the specified geographical area within the specified time period from the event information of the reported events;

[0140] An early warning module 53 for inputting event information of at least one target event into a risk early warning model, so as to perform risk identification on at least one risk matter by using at least one target risk identification capability of the risk early warning model, and obtain a risk early warning result of a target risk scenario, where the risk early warning result includes the target risk matters that occur in the target risk scenario, the event information of the events belonging to the target risk matters, and at least one risk category of the target risk matters;

[0141] A push module 54 for pushing the risk early warning result of the target risk scenario to the push object corresponding to the push object identifier.

[0142] Further optionally, the above device further includes: a display module for displaying a configuration page of the target risk scenario, where the configuration page includes a scenario definition configuration item, a policy configuration item, a data source configuration item, and an information push configuration item; a configuration module for, in response to a configuration operation on the scenario definition configuration item, configuring a scenario identifier of the target risk scenario, at least one target risk identification capability, and at least one risk matter, where the at least one target risk identification capability is selected from multiple risk identification capabilities possessed by the risk early warning model; in response to a configuration operation on the policy configuration item, configuring a specified time period and a specified geographical area of the target risk scenario; in response to a configuration operation on the data source configuration item, configuring at least one event source channel identifier of the target risk scenario; and in response to a configuration operation on the information push configuration item, configuring a push object identifier corresponding to the target risk scenario.

[0143] Further optionally, when the display module displays the configuration page of the target risk scenario, it is specifically configured to: display a risk scenario management page, where the risk scenario management page includes a page area of multiple risk scenarios; and in response to a trigger operation on a scenario configuration control in the page area of the target risk scenario, display the configuration page of the target risk scenario.

[0144] Further optionally, the display module is further configured to: after the target risk scenario is configured, in response to a trigger operation on a scenario details control in the page area of the target risk scenario, display a scenario details page of the target risk scenario; and in response to a modification operation triggered in the scenario details page, modify the configuration information of the target risk scenario in the scenario details page to obtain the modified configuration information of the target risk scenario.

[0145] Further optionally, the push module 54 is specifically configured to: perform clustering analysis on the event information of the events belonging to the target risk matter in the target risk scenario, generate early warning push content for the target risk scenario according to the clustering analysis result; generate an early warning push record for the target risk matter according to the event name of the target risk scenario, the number of target events belonging to the risk matter, at least one risk category, and the early warning push content; and push the early warning push records of the respective target risk matters in the target risk scenario to the push object corresponding to the push object identifier.

[0146] Further optionally, the risk early warning model includes multiple sub-models with different risk identification capabilities; correspondingly, the early warning module 53 is specifically configured to: input the event information of at least one target event into the risk early warning model, so that the sub-models corresponding to at least one target risk identification capability respectively perform risk identification; and generate a risk early warning result for the target risk scenario according to the risk identification results respectively output by the sub-models corresponding to at least one target risk identification capability.

[0147] Further optionally, the sub-models include at least two of the following:

[0148] A group hot spot research and judgment early warning model with group hot spot identification ability, which is used to identify whether the risk category of the risk matter is a group hot spot;

[0149] A sudden hot spot research and judgment early warning model with sudden hot spot identification ability, which is used to identify whether the risk category of the risk matter is a sudden hot spot;

[0150] A continuous hot spot research and judgment early warning model with continuous hot spot identification ability, which is used to identify whether the risk category of the risk matter is a continuous hot spot;

[0151] A repeated complaint research and judgment early warning model with repeated complaint identification ability, which is used to identify whether the risk category of the risk matter is a repeated complaint problem;

[0152] A high-frequency subject research and judgment early warning model with high-frequency subject identification ability, which is used to identify whether the risk category of the risk matter is a high-frequency subject;

[0153] A hot spot trend research and judgment early warning model with high-incidence problem identification ability, which is used to identify whether the public opinion risk category of the risk matter is a high-incidence problem;

[0154] A sensitive appeal research and judgment early warning model with sensitive appeal identification ability, which is used to identify whether the risk category of the risk matter is a sensitive appeal;

[0155] A concerned topic research and judgment early warning model with concerned topic identification ability, which is used to identify whether the risk category of the risk matter is a concerned topic;

[0156] A real-time hot-spot research, judgment and early warning model with real-time hot-spot recognition ability, which is used to identify whether the risk category of a risk matter is a real-time hot spot;

[0157] A repeated problem research, judgment and early warning model with repeated problem recognition ability, which is used to identify whether the risk category of a risk matter is a repeated problem;

[0158] A long-pending problem research, judgment and early warning model with long-pending problem recognition ability, which is used to identify whether the risk category of a risk matter is a long-pending problem.

[0159] Figure 5 The device shown can execute Figure 2 The method shown, and its implementation principle and technical effects will not be elaborated. For the Figure 5 device shown, the specific ways for each module and unit to execute operations have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0160] It should be noted that the execution subject of each step of the method provided in the above embodiments can be the same device, or the method can also be executed by different devices as the execution subject. For example, the execution subject of steps 201 to 204 can be device A; or, the execution subject of steps 201 and 202 can be device A, and the execution subject of steps 203 to 204 can be device B; and so on.

[0161] In addition, in some processes described in the above embodiments and the accompanying drawings, a plurality of operations appear in a specific order, but it should be clearly understood that these operations can be executed not in the order in which they appear in this article or in parallel. The operation numbers such as 201 and 202 are only used to distinguish different operations, and the numbers themselves do not represent any execution order. In addition, these processes can include more or fewer operations, and these operations can be executed in sequence or in parallel. It should be noted that the descriptions such as "first" and "second" in this article are used to distinguish different messages, devices, modules, etc., do not represent a sequence, and do not limit that "first" and "second" are of different types.

[0162] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data need to comply with relevant laws, regulations and standards of relevant countries and regions, and corresponding operation entrances are provided for users to choose to authorize or refuse.

[0163] Figure 6A schematic structural diagram of an electronic device provided by an embodiment of the present application. As Figure 6 shown, the electronic device includes: a memory 61 and a processor 62;

[0164] The memory 61 is used to store computer programs and can be configured to store various other data to support operations on a computing platform. Examples of such data include instructions for any application or method operating on the computing platform, contact data, phone book data, messages, pictures, videos, etc.

[0165] The memory 61 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read only memory (EEPROM), erasable programmable read only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, a magnetic disk or an optical disc.

[0166] The processor 62 is coupled to the memory 61 and is used to execute the computer program in the memory 61 for: performing the steps in the risk warning method.

[0167] Further optionally, as Figure 6 shown, the electronic device further includes: other components such as a communication component 63, a display 64, a power supply component 65, an audio component 66, etc. Figure 6 Only some components are schematically shown, and it does not mean that the electronic device only includes Figure 6 the components shown. Additionally, Figure 6 the components within the dashed box are optional components, not mandatory components, and can be determined according to the product form of the electronic device. The electronic device in this embodiment can be implemented as a terminal device such as a desktop computer, a laptop computer, a smart phone, or an IOT (Internet of Things) device, or can also be a server device such as a conventional server, a cloud server, or a server array. If the electronic device in this embodiment is implemented as a terminal device such as a desktop computer, a laptop computer, or a smart phone, it can include Figure 6 the components within the dashed box; if the electronic device in this embodiment is implemented as a server device such as a conventional server, a cloud server, or a server array, it may not include Figure 6 the components within the dashed box.

[0168] For the detailed implementation process of the processor to execute each action, reference may be made to the relevant descriptions in the foregoing method embodiments or device embodiments, which will not be elaborated herein.

[0169] Correspondingly, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, and when the computer program is executed, it can implement each step executable by an electronic device in the foregoing method embodiments.

[0170] Correspondingly, an embodiment of the present application further provides a computer program product, including a computer program / instructions, and when the computer program / instructions are executed by a processor, the processor can implement each step executable by an electronic device in the foregoing method embodiments.

[0171] The above communication component is configured to facilitate communication between the device where the communication component is located and other devices in a wired or wireless manner. The device where the communication component is located can access a wireless network based on a communication standard, such as WiFi (Wireless Fidelity), 2G (2 Generation), 3G (3 Generation), 4G (4 Generation) / LTE (long Term Evolution), 5G (5 Generation), etc. mobile communication networks, or a combination thereof. In an exemplary embodiment, the communication component receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component further includes a Near Field Communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on Radio Frequency Identification (RFID) technology, Infrared Data Association (IrDA) technology, Ultra Wide Band (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0172] The above display includes a screen, and the screen may include a Liquid Crystal Display (LCD) and a Touch Panel (TP). If the screen includes a touch panel, the screen can be implemented as a touch screen to receive input signals from a user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors can not only sense the boundaries of touch or swipe actions, but also detect the duration and pressure associated with the touch or swipe operations.

[0173] The above power supply component provides power for various components of the device where the power supply component is located. The power supply component may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for the device where the power supply component is located.

[0174] The above audio component can be configured to output and / or input audio signals. For example, the audio component includes a microphone (MIC). When the device where the audio component is located is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode, the microphone is configured to receive external audio signals. The received audio signals can be further stored in a memory or transmitted via a communication component. In some embodiments, the audio component further includes a speaker for outputting audio signals.

[0175] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) that contain computer-usable program code.

[0176] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0177] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device that implements the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0178] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, causing a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one process or a plurality of processes and / or blocks Figure 1 one process or a plurality of processes and / or blocks Figure 1 steps for implementing the functions specified in one block or a plurality of blocks.

[0179] In a typical configuration, a computing device includes one or more processors (Central Processing Unit, CPU), an input / output interface, a network interface, and memory.

[0180] The memory may include non-permanent memory in the computer-readable medium, in the form of random access memory (Random Access Memory, RAM) and / or non-volatile memory such as read only memory (Read Only Memory, ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0181] The computer-readable medium includes permanent and non-permanent, removable and non-removable media which can store information by any method or technology. The information may be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change random access memory (Phase Change RAM, PRAM), static random access memory (Static Random-Access Memory, SRAM), dynamic random access memory (Dynamic Random Access Memory, DRAM), other types of random access memory (Random Access Memory, RAM), read only memory (Read Only Memory, ROM), electrically erasable programmable read only memory (Electrically-Erasable Programmable Read-Only Memory, EEPROM), flash memory or other memory technologies, compact disc read only memory (CD-ROM), digital versatile disc (Digital versatile disc, DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0182] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, commodity or device comprising a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, commodity or device comprising the element.

[0183] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the scope of the claims of the present application.

Claims

1. A risk warning method, characterized in that, Including: Obtain the configuration information of the target risk scenario, where the configuration information includes: identifiers of at least one event source channel, a specified time period, a specified geographical area, at least one risk matter, at least one target risk identification capability, and a push object identifier. Different target risk identification capabilities are used to identify different risk categories of risk matters; According to the identifiers of at least one event source channel, obtain the event information of the events reported by at least one event source channel, and obtain the event information of at least one target event that occurred in the specified geographical area during the specified time period from the event information of the reported events; Input the event information of the at least one target event into a risk warning model, so as to use the at least one target risk identification capability through the risk warning model to perform risk identification for at least one risk matter, and obtain the risk warning result of the target risk scenario. The risk warning result includes the target risk matters that appear in the target risk scenario, the event information of the events belonging to the target risk matters, and at least one risk category of the target risk matters; Push the risk warning result of the target risk scenario to the push object corresponding to the push object identifier.

2. The method according to claim 1, wherein Before obtaining the configuration information of the target risk scenario, it further includes: Display the configuration page of the target risk scenario, where the configuration page includes a scenario definition configuration item, a policy configuration item, a data source configuration item, and an information push configuration item; In response to a configuration operation for the scenario definition configuration item, configure the scenario identifier, at least one target risk identification capability, and at least one risk matter of the target risk scenario, where at least one target risk identification capability is selected from multiple risk identification capabilities possessed by the risk warning model; In response to a configuration operation for the policy configuration item, configure the specified time period and the specified geographical area of the target risk scenario; In response to a configuration operation for the data source configuration item, configure at least one event source channel identifier of the target risk scenario; In response to a configuration operation for the information push configuration item, configure the push object identifier corresponding to the target risk scenario.

3. The method according to claim 2, characterized in that, Displaying the configuration page of the target risk scenario includes: Display a risk scenario management page, where the risk scenario management page includes a page area for multiple risk scenarios; In response to a trigger operation on a scenario configuration control in the page area of the target risk scenario, display the configuration page of the target risk scenario.

4. The method according to claim 3, characterized in that, It further includes: After the target risk scenario is configured, in response to a trigger operation on a scenario details control in the page area of the target risk scenario, display the scenario details page of the target risk scenario; In response to a modification operation triggered in the scenario details page, modify the configuration information of the target risk scenario in the scenario details page to obtain the modified configuration information of the target risk scenario.

5. The method according to any one of claims 1 to 4, characterized in that Pushing the risk warning result of the target risk scenario to the push object corresponding to the push object identifier includes: Cluster analyze the event information of the events belonging to the target risk matter for the target risk matter that appears in the target risk scenario, and generate the early warning push content of the target risk scenario according to the cluster analysis result; Generate an early warning push record of the target risk matter according to the event name of the target risk scenario, the number of target events belonging to the risk matter, at least one risk category, and the early warning push content; Push the early warning push records of each target risk matter in the target risk scenario to the push object corresponding to the push object identifier.

6. The method according to claim 1, characterized in that, The risk early warning model includes multiple sub-models with different risk identification capabilities; Correspondingly, input the event information of the at least one target event into the risk early warning model, so as to use at least one target risk identification capability by the risk early warning model to perform risk identification for at least one risk matter, and obtain the risk early warning result of the target risk scenario, including: Input the event information of the at least one target event into the risk early warning model, so that the sub-models corresponding to at least one target risk identification capability respectively perform risk identification; Generate the risk early warning result of the target risk scenario according to the risk identification results output by the sub-models corresponding to at least one target risk identification capability respectively.

7. The method according to claim 6, characterized in that, The sub-model includes at least two of the following: A group hot spot research and judgment early warning model with group hot spot identification ability, which is used to identify whether the risk category of the risk matter is a group hot spot; A sudden hot spot research and judgment early warning model with sudden hot spot identification ability, which is used to identify whether the risk category of the risk matter is a sudden hot spot; A continuous hot spot research and judgment early warning model with continuous hot spot identification ability, which is used to identify whether the risk category of the risk matter is a continuous hot spot; A repeated complaint research and judgment early warning model with repeated complaint identification ability, which is used to identify whether the risk category of the risk matter is a repeated complaint problem; A high-frequency subject research and judgment early warning model with high-frequency subject identification ability, which is used to identify whether the risk category of the risk matter is a high-frequency subject; A hot spot trend research and judgment early warning model with high-incidence problem identification ability, which is used to identify whether the risk category of the risk matter is a high-incidence problem; A sensitive appeal research and judgment early warning model with sensitive appeal identification ability, which is used to identify whether the risk category of the risk matter is a sensitive appeal; A concerned topic research and judgment early warning model with concerned topic identification ability, which is used to identify whether the risk category of the risk matter is a concerned topic; A real-time hot spot research and judgment early warning model with real-time hot spot identification ability, which is used to identify whether the risk category of the risk matter is a real-time hot spot; A repeated attack problem research and judgment early warning model with repeated attack problem identification ability, which is used to identify whether the risk category of the risk matter is a repeated attack problem; A long-outstanding problem research and judgment early warning model with long-outstanding problem identification ability, which is used to identify whether the risk category of the risk matter is a long-outstanding problem.

8. A risk warning method, characterized in that, Applied to a cloud server, the method includes: In response to a configuration operation on a configuration page that displays a target risk scenario in a terminal device, obtain configuration information of the target risk scenario, where the configuration information includes: identifiers of at least one event source channel, a specified time period, a specified geographical area, at least one risk matter, at least one target risk identification capability, and a push object identifier, and different target risk identification capabilities are used to identify different risk categories of risk matters; According to the identifiers of at least one event source channel, obtain event information of events reported by at least one event source channel, and obtain event information of at least one target event that occurred in the specified geographical area within the specified time period from the event information of the reported events; Input the event information of the at least one target event into a risk early warning model, so that the risk early warning model uses the at least one target risk identification capability to perform risk identification on at least one risk matter, and obtain a risk early warning result of the target risk scenario, where the risk early warning result includes the target risk matters that appear in the target risk scenario, the event information of the events belonging to the target risk matters, and at least one risk category of the target risk matters; Push the risk early warning result of the target risk scenario to the push object corresponding to the push object identifier.

9. An electronic device, characterized in that, Comprising: A memory and a processor; The memory is used to store a computer program; The processor is coupled to the memory and is used to execute the computer program to execute the steps in the method according to any one of claims 1-8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it causes the processor to be able to implement the steps in the method according to any one of claims 1-8.

Images