Emergency report generation method, device, equipment and storage medium

Abstract

The present invention discloses a method, device, equipment and storage medium for generating an emergency report, belonging to the field of data processing technology. The present invention obtains raw data corresponding to a target event and performs preprocessing to obtain first preprocessed data; the first preprocessed data is input into a preset event recognition model to identify the event type and event progress stage of the target event; based on the event type and the event progress stage, a corresponding target template is determined from a preset report template library; based on a preset data screening model, the contribution of different data in the first preprocessed data to the event level is calculated to determine key data; based on a natural language processing algorithm, the key data is converted into descriptive text, filled in the target template, and an emergency report is obtained. The present invention improves the readability of emergency reports as a whole and is applicable to the generation of reports for various emergency events.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method, device, equipment and storage medium for generating an emergency report. Background Art

[0002] In the event of a sudden mass incident (such as a natural disaster, a major traffic accident, or a public health incident), emergency management departments, medical institutions, and related organizations at all levels often need to respond quickly to ensure effective rescue and incident handling. During this process, timely and accurate emergency reports are required, including key information such as casualties, the progress of medical treatment, and resource sourcing. However, in related technologies, automatic report generation systems can only generate reports for a single type of event, and the generated reports are usually structured text, which cannot fully reflect the logical relationships between data and is not conducive to a comprehensive understanding of the overall context of the incident. Furthermore, data information is redundant, and the report content lacks focus, resulting in poor readability.

[0003] Therefore, in the process of generating event reports, the related technology has technical problems such as a single report type and poor readability of the report content. Summary of the Invention

[0004] The main purpose of the present invention is to provide a method, device, equipment and storage medium for generating an emergency report, aiming to solve the technical problems in the related art that the report type is single and the report content is poorly readable.

[0005] To achieve the above object, the present invention provides a method for generating an emergency report, the method comprising the following steps:

[0006] S1, obtaining raw data corresponding to a target event and preprocessing it to obtain first preprocessed data;

[0007] S2, inputting the first pre-processed data into a preset event recognition model to identify the event type and event progress stage of the target event;

[0008] S3, based on the event type and the event progress stage, determining the corresponding target template from the preset report template library;

[0009] S4, based on a preset data screening model, calculating the contribution of different data in the first preprocessed data to the event level, and determining key data;

[0010] S5, based on natural language processing algorithms, converts key data into descriptive text, fills the target template, and obtains the emergency report.

[0011] Optionally, in step S4, the steps of constructing the preset data screening model specifically include:

[0012] Acquire first historical emergency event data and perform preprocessing to obtain a first training set; the first historical emergency event data at least includes event level, event occurrence time, event scope, event type, total number of people affected by the event, number of casualties, number of first-level triage patients, number of second-level triage patients, and resource consumption;

[0013] The first training set is input into the decision tree model, and the model is trained based on the dynamic weighted information gain formula to obtain a preset data screening model; the dynamic weighted information gain formula is: ;

[0014] in, S is the first training set, is the dynamic information gain of the first training set S, Characterized by i The association weight with the event level at time t, Based on characteristics i After partitioning, the subset is the size of the first training set S, For subset The size of The first training set S The entropy of For subset Entropy;

[0015] Association weight The calculation formula is: ;

[0016] in, for t The duration of the event corresponding to the moment, for t The resource consumption corresponding to the moment, Characterized by i The adjustment factor for the duration of the event, Features i Adjustment coefficient on resource consumption. Optionally, in step S2, the steps of constructing the preset event recognition model specifically include:

[0017] Acquire the second historical emergency event data, perform preprocessing on the data, and obtain multiple historical emergency event reports;

[0018] For each historical emergency report, the event type and event progress stage are annotated to obtain the second training set; the event type annotation includes at least natural disasters, traffic accidents and public health events, and the event progress stage annotation includes at least the early stage, middle stage and late stage of the event;

[0019] Extract relevant features of each historical emergency event report; relevant features include at least the time of occurrence, duration of the event, location of the event, total number of people affected by the event, number of casualties, number of first-level triage patients, number of second-level triage patients, and resource consumption;

[0020] The second training set is input into the deep learning model, and training is performed based on relevant features to obtain a preset event recognition model.

[0021] Optionally, the target template has a specific report framework, and step S5 specifically includes:

[0022] Based on natural language processing algorithms, identify the mapping relationship between key data and fields in predefined sentence templates;

[0023] Based on the mapping relationship, key data is filled into the corresponding fields of the predefined sentence template to obtain descriptive text;

[0024] Based on the reporting framework, descriptive text is inserted into the target template to obtain the incident report.

[0025] Optionally, step S1 specifically includes:

[0026] Extract the original data corresponding to the target event to obtain plain text data;

[0027] The plain text data is standardized and completed to obtain first preprocessed data.

[0028] Optionally, after step S5, the method further includes:

[0029] a1. Based on the file format requirements of the corresponding receiving department of the emergency report, the emergency report is converted into a target format report that meets the file format requirements;

[0030] a2, in response to the user's one-click reporting instruction, select the reporting channel corresponding to the receiving department;

[0031] a3. Based on the reporting channel, upload the target format report to the receiving department.

[0032] Optionally, after step a3, the method further includes:

[0033] a4, obtaining and monitoring the processing progress of the target format report based on a preset time interval;

[0034] a5. If the processing progress changes, the current processing progress is recorded in chronological order to form a historical processing record arranged in chronological order.

[0035] In addition, to achieve the above-mentioned purpose, the present invention also provides an emergency report generating device, the device comprising:

[0036] A data preprocessing module is used to obtain the original data corresponding to the target event and perform preprocessing to obtain first preprocessed data;

[0037] An event recognition module, configured to input the first pre-processed data into a preset event recognition model to identify the event type and event progress stage of the target event;

[0038] A template selection module is used to determine a corresponding target template from a preset report template library based on the event type and the event progress stage;

[0039] A data screening module, configured to calculate the contribution of different data in the first pre-processed data to the event level based on a preset data screening model, and determine key data;

[0040] The report generation module is used to convert key data into descriptive text based on natural language processing algorithms, fill in the target template, and obtain emergency event reports.

[0041] Furthermore, to achieve the above-mentioned purpose, the present invention also provides an emergency event report generation device, which includes: a memory, a processor, and an emergency event report generation program stored in the memory and executable on the processor, wherein the emergency event report generation program is configured to implement the steps of the emergency event report generation method as described above.

[0042] Furthermore, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, on which an emergency event report generation program is stored. When the emergency event report generation program is executed by a processor, the steps of the emergency event report generation method as described above are implemented.

[0043] The present invention presets different report templates for different event types and different event progress stages, constructs a preset template library, and can realize the generation of event reports for a variety of emergencies. Secondly, according to the contribution of different data to the event level, key data is screened out, which can avoid data redundancy in the report content and make the information focus more prominent. On this basis, after using a natural language processing algorithm to convert key data in the form of structured text into descriptive text, the template is filled in, so that the logic between the information elements in the emergency report content is clear, which helps the report receiving department to accurately understand the entire event. Therefore, the present invention improves the readability of emergency reports as a whole. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a structural diagram of an emergency event report generating device in a hardware operating environment involved in an embodiment of the present invention;

[0045] Figure 2This is a flow chart of a first embodiment of a method for generating an emergency report according to the present invention;

[0046] Figure 3 This is a flow chart of a second embodiment of a method for generating an emergency report according to the present invention;

[0047] Figure 4 This is a flow chart of the third embodiment of the method for generating an emergency report according to the present invention.

[0048] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0049] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0050] The inventive concept of the present application is further described below with reference to some specific embodiments and implementation methods.

[0051] Reference Figure 1 , Figure 1 The present invention provides a schematic diagram of an emergency event report generation device structure for a hardware operating environment according to an embodiment of the present invention.

[0052] like Figure 1 As shown, the emergency report generation device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display and an input unit, such as a keyboard. Optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a wireless fidelity (WI-FI) interface). The memory 1005 may be a high-speed random access memory (RAM) or a stable non-volatile memory (NVM), such as a disk drive. The memory 1005 may also be a storage device independent of the processor 1001.

[0053] Those skilled in the art will understand that Figure 1 The structure shown in the figure does not constitute a limitation to the emergency report generating device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0054] like Figure 1 As shown, the memory 1005 as a computer-readable storage medium may include an operating system, a network communication module, a user interface module, and an emergency report generation program.

[0055] exist Figure 1 In the emergency report generation device shown, the network interface 1004 is mainly used for data communication with other devices; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the emergency report generation device of the present invention can be set in the emergency report generation device, and the emergency report generation device calls the emergency report generation program stored in the memory 1005 through the processor 1001, and executes the emergency report generation method provided by the embodiment of the present invention.

[0056] The embodiment of the present invention provides a method for generating an emergency report, referring to Figure 2 , Figure 2 The figure is a flow chart of a first embodiment of a method for generating an emergency report according to the present invention.

[0057] In this embodiment, the method for generating an emergency report includes:

[0058] Step S1: Obtain original data corresponding to a target event and perform preprocessing to obtain first preprocessed data.

[0059] In step S1, the pre-processing steps specifically include:

[0060] Step S11: extracting the original data corresponding to the target event to obtain plain text data.

[0061] Step S12: performing standardization processing and data completion processing on the plain text data to obtain first preprocessed data.

[0062] In this embodiment, target events include, but are not limited to, mass emergencies such as traffic accidents, natural disasters, and public health incidents. In specific implementations, after a target event occurs, the emergency report generation device can obtain raw data from multiple data sources. These data sources include, but are not limited to, hospital information systems, emergency management platforms, sensors, and monitoring equipment.

[0063] Specifically, raw data can be in various formats, such as tables, statistical charts, and plain text. First, a specific data extraction method is used to extract raw data from various data sources to obtain plain text data that is easy to process. This data extraction method can use one or more of OCR technology, chart data extraction software, and machine learning methods.

[0064] Standardize the obtained plain text data. For example, convert time or temperature data in different units into a standard format with unified units; for another example, scale data in different dimensions to a unified numerical range for data processing and analysis.

[0065] If there are missing data in the normalized plain text data, a preset data completion algorithm is used to complete the missing data. Specifically, the preset data completion algorithm can use one or more of interpolation, mean filling, median filling, KNN filling, and multiple interpolation methods.

[0066] Step S2: Input the first pre-processed data into a preset event recognition model to identify the event type and event progress stage of the target event.

[0067] In step S2, the steps of constructing the preset event recognition model specifically include:

[0068] Step S21: Acquire second historical emergency event data, and perform preprocessing to obtain multiple historical emergency event reports.

[0069] Specifically, the secondary historical emergency data covers different types of emergencies and, to the extent possible, includes various characteristics of the events (such as the time of occurrence, duration, location, total number of people affected, casualties, number of first-level triage visits, number of second-level triage visits, and resource consumption). Preprocessing includes, but is not limited to, denoising, standardization, and missing value processing.

[0070] Step S22: labeling the event type and the event progress stage for each historical emergency event report to obtain a second training set.

[0071] Among them, the event type labeling includes at least natural disasters, traffic accidents and public health events, and the event progress stage labeling includes at least the early stage of the event, the middle stage of the event and the late stage of the event.

[0072] Annotation can be performed manually or semi-automatically. Manual annotation requires annotators to categorize events based on their descriptions, but ensures high accuracy. Semi-automatic annotation uses natural language processing (NLP) technology to extract key information and perform preliminary annotations, which are then manually reviewed.

[0073] Finally, the labeled data will be used as the second training set. The historical emergency reports in the second training set contain event type labels and event progress stage labels.

[0074] Step S23: extract relevant features of each historical emergency report.

[0075] Among them, the relevant characteristics include at least the time of the incident, the duration of the incident, the location of the incident, the total number of people affected by the incident, the number of casualties, the number of people in the first-level triage, the number of people in the second-level triage, and resource consumption.

[0076] Step S24: Input the second training set into the deep learning model, perform training based on relevant features, and obtain a preset event recognition model.

[0077] Specifically, for each historical emergency report, relevant features associated with the event type and stage of the incident were extracted, including the time of occurrence, duration, location, total number of people affected, casualties, number of first-level triage patients, number of second-level triage patients, and resource consumption. The second training set was input into the deep learning model, which was trained to learn the mapping relationship between relevant features and event types and stages of the incident. After multiple rounds of iteration, a preset event recognition model was obtained.

[0078] Finally, the preset event recognition model is used to identify the event type and event progress stage corresponding to the target event from the first preprocessed data.

[0079] Step S3: Based on the event type and the event progress stage, a corresponding target template is determined from a preset report template library.

[0080] In this embodiment, a preset template library is pre-built to meet the needs of emergency event report generation in various scenarios. Specifically, corresponding report templates are preset for different event types and different event progress stages and stored in the preset template library. During implementation, the preset template library can first be searched for multiple report templates corresponding to the event type based on the event type. Then, further search can be performed based on the event progress stage to obtain the target template corresponding to the event progress stage from the multiple report templates corresponding to the event type.

[0081] In some specific embodiments, the query process can be implemented by function 1:

[0082] Function 1: ;

[0083] in, is the event type, The event progress stage, is the target template, Indicates a query operation. Specifically, the event type and the stage of event progression Input function 1 as known data, and then execute the query operation involved in function 1 according to the known input conditions to obtain the target template .

[0084] For example, you can preset initial report templates, follow-up report templates, and final report templates for different stages of the same emergency. The initial report template can focus on reporting the incident overview, the follow-up report template can focus on reporting the consumption of medical resources and the progress of treatment, and the final report template can focus on reporting the summary and analysis of the incident.

[0085] Step S4: Based on a preset data screening model, the contribution of different data in the first preprocessed data to the event level is calculated to determine the key data.

[0086] Specifically, the pre-processed data is first input into a preset data screening model to calculate the contribution of different data to the event level. Based on the contribution, the different data are ranked by contribution, and data with a contribution greater than a first preset value is determined as key data.

[0087] In step S4, the steps of constructing the preset data screening model specifically include:

[0088] Step S41: Acquire first historical emergency event data and perform preprocessing to obtain a first training set.

[0089] Among them, the first historical emergency event data at least includes the event level, event occurrence time, event scope, event type, total number of people affected by the event, number of casualties, number of first-level triage patients, number of second-level triage patients, and resource consumption.

[0090] Preprocessing of the first historical emergency event data includes, but is not limited to, missing value processing, standardization, and categorical feature encoding to obtain a first training set. Categorical features include, but are not limited to, event time, event scope, event type, total number of people affected, number of casualties, number of first-level triage patients, number of second-level triage patients, and resource consumption.

[0091] Step S42: Input the first training set into the decision tree model, perform model training based on the dynamic weighted information gain formula, and obtain a preset data screening model.

[0092] Among them, the dynamic weighted information gain formula is: ;in, S is the first training set, is the dynamic information gain of the first training set S, Characterized by i The association weight with the event level at time t, Based on characteristics i After partitioning, the subset The first training set S The size of For subset The size of The first training set S The entropy of For subset entropy.

[0093] Association weight The calculation formula is: ;in, for t The duration of the event corresponding to the moment, for t The resource consumption corresponding to the moment, Characterized by i The adjustment factor for the duration of the event, Features i Adjustment factor on resource consumption.

[0094] Specifically, the decision tree model repeatedly performs feature selection and splitting through the first training set, and each time selects the feature with the largest information gain for splitting according to the information gain calculation result, until a complete tree structure is obtained, and finally a preset data screening model is obtained.

[0095] It is worth mentioning that in the process of decision tree training, the information gain formula is improved by introducing dynamic weighting. i Based on the original information gain, the duration of the event and the amount of resource consumption are comprehensively considered to determine the feature i The influence of contribution is further dynamically weighted to obtain the characteristics i The dynamically weighted information gain value at time t. This allows the trained preset data screening model to filter out the most important key data for the current event stage. Then, based on this key data, an emergency report is generated, avoiding information redundancy in the emergency report and making the report more focused and readable.

[0096] Among them, the adjustment coefficient for t At the moment, characteristics i The rate of change as the duration of the event increases. Adjustment factor for t At the moment, characteristics i The rate of change as resource consumption increases.

[0097] Step S5: Based on the natural language processing algorithm, the key data is converted into descriptive text, which is then filled into the target template to obtain the emergency report.

[0098] Step S5 specifically includes:

[0099] Step S51: Based on a natural language processing algorithm, identify the mapping relationship between key data and fields in a predefined sentence template.

[0100] Step S52: Based on the mapping relationship, key data is filled into corresponding fields of the predefined sentence template to obtain descriptive text.

[0101] Step S53: Based on the report framework, the descriptive text is inserted into the target template to obtain an emergency report.

[0102] Specifically, the natural language processing algorithm is used to first pre-process the input key data through word segmentation, part-of-speech tagging, named entity recognition, etc., and then the machine learning model is used to identify the mapping relationship between the key data and the fields of the predefined sentence template.

[0103] For example, if the key data is "Casualties: 50; Seriously injured: 15; Flesh wound: 35," the corresponding predefined sentence template 1 is "In this accident, a total of (field 1) people were injured, of whom (field 2) were seriously injured, and (field 3) were slightly injured." The identified mapping relationship is: data "50" is mapped to "Field 1," data "15" is mapped to "Field 2," and data "35" is mapped to "Field 3."

[0104] Based on the identified mapping relationship, the key data is filled into the corresponding fields, and the complete descriptive text in the above example is obtained as "In this accident, a total of 50 people were injured, of which 15 were seriously injured and 35 were slightly injured."

[0105] It is worth mentioning that the target template has a specific report framework, which determines the structure and filling rules of the emergency report. According to the structure and filling rules of the report, the descriptive text is inserted into the corresponding position of the target template to obtain the emergency report.

[0106] For example, the target template's report framework consists of structural layers such as the event summary, the cause of the incident, and emergency measures. The filling rules specify the mapping relationship between predefined sentence templates and structural layers (e.g., predefined sentence template 1 maps to the event summary). Based on this mapping relationship, the descriptive text generated based on the predefined sentence templates is inserted into the corresponding structural layer (e.g., the descriptive text generated based on predefined sentence template 1 is inserted into the event summary).

[0107] In some other specific embodiments, the key data filling process can be implemented by function 2:

[0108] Function 2: ;in, Target template, To preset natural language generation model, For key data, The text of the generated incident report. The template is filled.

[0109] Specifically, function 2 first executes the key data Fill in target template to obtain a filled emergency report; and then perform a step of optimizing the content of the emergency report using a preset natural language generation model to obtain a fluent and natural final report.

[0110] In this embodiment, a variety of report templates are preset and a preset template library is constructed according to different event types and event progress stages, so that event reports for various emergencies can be generated. At the same time, key data is screened out based on the contribution of different data to the event level, avoiding data redundancy in the report content and making the focus of the information more prominent. On this basis, through a natural language processing algorithm, the key data in the form of structured text is converted into descriptive text and filled into the target template, so that the information elements in the emergency report are logically clear, making it easier for the receiving department to accurately understand the overall situation of the event. Therefore, the present invention improves the readability of the emergency report as a whole. Furthermore, it helps relevant departments to deal with emergencies more efficiently and make scientific decisions.

[0111] Furthermore, a second embodiment is proposed based on the first embodiment, referring to Figure 3 , Figure 3 This is a flow chart of the second embodiment of the method for generating an emergency report according to the present invention.

[0112] In this embodiment, after step S5, the method further includes:

[0113] Step a1: Based on the file format requirements of the corresponding receiving department of the emergency report, the emergency report is converted into a target format to obtain a report that meets the file format requirements.

[0114] Step a2: In response to the user's one-click reporting instruction, select the reporting channel corresponding to the receiving department.

[0115] Step a3: Based on the reporting channel, upload the target format report to the receiving department.

[0116] Specifically, you first need to obtain the receiving department's file format requirements for emergency reports, including file type (PDF, Word, Excel, etc.), document structure (such as whether specific headers, field names, and timestamp formats are required), and supported attachment types.

[0117] In the specific implementation process, the format requirements can be extracted from the document template, interface document or configuration file obtained from the receiving department. When performing format conversion, the corresponding document conversion tool or document conversion service can be directly called to complete it.

[0118] In some specific embodiments, the upload process can be implemented by function three:

[0119] Function three: ;in, For current emergency reports, is the target format, For the converted emergency report, Represents a conversion operation.

[0120] Specifically, report the current emergency and target format As the known data is input into function three, then according to the known conditions of the input, the corresponding document conversion tool or document conversion service is called through function three to complete the conversion operation and obtain the converted emergency report. .

[0121] Furthermore, in this embodiment, users can quickly upload emergency reports with a single click. Specifically, when a user issues a "one-click report" command, the appropriate reporting channel is automatically selected based on the receiving department. Specifically, the appropriate reporting channel can be selected based on the receiving department's configuration file. In practice, the reporting channel can be one or more of email, cloud storage, and FTP.

[0122] In some specific embodiments, the upload process can be implemented by function four:

[0123] Function 4: ;

[0124] in, For the converted emergency report, For reporting channels, Indicates an upload operation.

[0125] Specifically, the converted emergency report and reporting channels Input function four as known data, and then execute the upload operation involved in function four according to the known input conditions to complete the upload task of the emergency report.

[0126] During the emergency report upload process, to ensure the visibility of the emergency report upload results, you can confirm whether the file has been successfully uploaded based on the upload results returned by the reporting channel, and provide feedback to the user (such as success prompts, failure retry prompts, etc.).

[0127] In this embodiment, emergency reports are converted to meet the file format requirements of the receiving department, ensuring that the report format is highly consistent with departmental requirements. Furthermore, the system automatically selects the appropriate reporting channel in response to the user's one-click reporting command, streamlining the user process, improving work efficiency, and reducing manual intervention. This not only enhances the accuracy and timeliness of report delivery, but also improves the user experience and saves time and resources.

[0128] Furthermore, a third embodiment is proposed based on the second embodiment, referring to Figure 4 , Figure 4 This is a flow chart of the third embodiment of the method for generating an emergency report according to the present invention.

[0129] In this embodiment, after step a3, the method further includes:

[0130] Step a4: Based on a preset time interval, obtain and monitor the processing progress of the target format report.

[0131] Step a5: If the processing progress changes, the current processing progress is recorded in chronological order to form a historical processing record arranged in chronological order.

[0132] Specifically, during the target format report upload process, the current status of the target format report and related log information are continuously obtained through the upload channel at preset intervals, from which processing progress information is extracted. Once a change in processing progress is detected, the current progress information is recorded and stored in chronological order to form a historical processing record. This progress information may include upload progress, process status, report time, and other information.

[0133] In addition, in some specific implementations, to facilitate subsequent analysis and tracking, a query interface may be provided, through which users can query by dimensions such as time, event type, and upload progress.

[0134] In this embodiment, by setting predetermined time intervals, the processing progress of target format reports can be continuously monitored and tracked in a timely manner. If the processing progress changes, the system automatically records the latest processing status in chronological order, forming a clear, chronological historical processing record. This not only provides real-time tracking of processing progress, but also provides detailed data support for subsequent analysis and decision-making. This combination of dynamic monitoring and historical recording can significantly improve the transparency and efficiency of emergency reporting.

[0135] Furthermore, to achieve the above-mentioned purpose, the present invention also provides an emergency event report generation device, which may include:

[0136] A data preprocessing module is used to obtain the original data corresponding to the target event and perform preprocessing to obtain first preprocessed data;

[0137] An event recognition module, configured to input the first pre-processed data into a preset event recognition model to identify the event type and event progress stage of the target event;

[0138] A template selection module is used to determine a corresponding target template from a preset report template library based on the event type and the event progress stage;

[0139] A data screening module, configured to calculate the contribution of different data in the first pre-processed data to the event level based on a preset data screening model, and determine key data;

[0140] The report generation module is used to convert key data into descriptive text based on natural language processing algorithms, fill in the target template, and obtain emergency event reports.

[0141] It should be noted that the functions that can be realized by each module in the emergency report generation device provided in this embodiment and the corresponding technical effects achieved can refer to the description of the specific implementation methods in each embodiment of the emergency report generation method of the present invention. For the sake of brevity of the description, they will not be repeated here.

[0142] In addition, an embodiment of the present invention further proposes a computer-readable storage medium, on which an emergency report generation program is stored, and when the emergency report generation program is executed by a processor, the steps of the emergency report generation method as described above are implemented. Therefore, no further description will be given here. In addition, the description of the beneficial effects of adopting the same method will not be repeated. For technical details not disclosed in the embodiment of the computer-readable storage medium involved in the present invention, please refer to the description of the embodiment of the method of the present invention. As an example, the program instructions can be deployed to be executed on one computing device, or on multiple computing devices located at one location, or on multiple computing devices distributed at multiple locations and interconnected by a communication network.

[0143] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0144] The above are only preferred embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for generating an emergency report, characterized in that: The method comprises: S1, obtaining raw data corresponding to a target event and preprocessing it to obtain first preprocessed data; S2, inputting the first pre-processed data into a preset event recognition model to identify the event type and event progress stage of the target event; S3, based on the event type and the event progress stage, determining a corresponding target template from a preset report template library; S4, calculating the contribution of different data in the first pre-processed data to the event level based on a preset data screening model, and determining key data; In S4, the steps of constructing the preset data screening model specifically include: Acquire first historical emergency event data and perform preprocessing to obtain a first training set; the first historical emergency event data includes at least the event level, event occurrence time, event scope, event type, total number of people affected by the event, number of casualties, number of first-level triage patients, number of second-level triage patients, and resource consumption; The first training set is input into the decision tree model, and the model is trained based on the dynamic weighted information gain formula to obtain a preset data screening model; the dynamic weighted information gain formula is: Among them, S is the first training set, InformationGain dynamic (S) is the dynamic information gain of the first training set S, w i (t) is the association weight between feature i and event level at time t, S i is the subset divided according to feature i, |S| is the size of the first training set S, |S i | is a subset S i The size of Entropy(S) is the entropy of the first training set S, Entropy(S i ) is a subset S i Entropy; The association weight w i The calculation formula for (t) is: w i (t)=a i ·EventDuration(t)+β i ·ResourceConsumption(t); Among them, EventDuration(t) is the duration of the event corresponding to time t, ResourceConsumption(t) is the resource consumption corresponding to time t, α i is the adjustment coefficient of feature i on the duration of the event, β i is the adjustment coefficient of feature i on resource consumption; S5, based on a natural language processing algorithm, converting the key data into descriptive text, filling the target template, and obtaining an emergency report.

2. The method for generating an emergency report according to claim 1, wherein: In S2, the steps of constructing the preset event recognition model specifically include: Acquire the second historical emergency event data, perform preprocessing on the data, and obtain multiple historical emergency event reports; For each historical emergency report, an event type labeling and an event progression labeling are performed to obtain a second training set; the event type labeling includes at least natural disasters, traffic accidents, and public health events, and the event progression labeling includes at least early event, mid-event, and late event; Extract relevant features of each historical emergency event report; the relevant features include at least the time of the event, the duration of the event, the location of the event, the total number of people affected by the event, the number of casualties, the number of people triaged at the first level, the number of people triaged at the second level, and the amount of resources consumed; The second training set is input into a deep learning model, and training is performed based on the relevant features to obtain the preset event recognition model.

3. The method for generating an emergency report according to claim 1, wherein: The target template has a specific reporting framework, and S5 specifically includes: Based on the natural language processing algorithm, identifying a mapping relationship between the key data and fields in a predefined sentence template; Based on the mapping relationship, the key data is filled into the corresponding field of the predefined sentence template to obtain the descriptive text; Based on the report framework, the descriptive text is inserted into the target template to obtain the emergency report.

4. The method for generating an emergency report according to claim 1, wherein: Said S1 specifically includes: Extracting the original data corresponding to the target event to obtain plain text data; The plain text data is subjected to standardization processing and data completion processing to obtain the first preprocessed data.

5. The method for generating an emergency report according to claim 1, wherein: After S5, the method further includes: a1, based on the file format requirements of the corresponding receiving department of the emergency report, convert the format of the emergency report to obtain a target format report that meets the file format requirements; a2, in response to the user's one-key reporting instruction, selecting the reporting channel corresponding to the receiving department; a3. Upload the target format report to the receiving department based on the reporting channel.

6. The method for generating an emergency report according to claim 5, wherein: After step a3, the method further includes: a4, obtaining and monitoring the processing progress of the target format report based on a preset time interval; a5. If the processing progress changes, the current processing progress is recorded in chronological order to form a historical processing record arranged in chronological order.

7. An emergency report generating device, characterized in that: The device comprises: A data preprocessing module is used to obtain the original data corresponding to the target event and perform preprocessing to obtain first preprocessed data; An event recognition module, configured to input the first pre-processed data into a preset event recognition model to identify the event type and event progress stage of the target event; A template selection module, configured to determine a corresponding target template from a preset report template library based on the event type and the event progress stage; A data screening module, configured to calculate the contribution of different data in the first pre-processed data to the event level based on a preset data screening model, and determine key data; The construction process of the preset data screening model is as follows: Acquire first historical emergency event data and perform preprocessing to obtain a first training set; the first historical emergency event data includes at least the event level, event occurrence time, event scope, event type, total number of people affected by the event, number of casualties, number of first-level triage patients, number of second-level triage patients, and resource consumption; The first training set is input into the decision tree model, and the model is trained based on the dynamic weighted information gain formula to obtain a preset data screening model; the dynamic weighted information gain formula is: Among them, S is the first training set, InformationGain dynamic (S) is the dynamic information gain of the first training set S, w i (t) is the association weight between feature i and event level at time t, S i is the subset divided according to feature i, |S| is the size of the first training set S, |S i | is a subset S i The size of Entropy(S) is the entropy of the first training set S, Entropy(S i ) is a subset S i Entropy; The association weight w i The calculation formula for (t) is: w i (t)=a i ·EventDuration(t)+β i ·ResourceConsumption(t); Among them, EventDuration(t) is the duration of the event corresponding to time t, ResourceConsumption(t) is the resource consumption corresponding to time t, α i is the adjustment coefficient of feature i on the duration of the event, β i is the adjustment coefficient of feature i on resource consumption; The report generation module is used to convert the key data into descriptive text based on a natural language processing algorithm, fill in the target template, and obtain an emergency report.

8. An emergency report generating device, characterized in that: The device includes: a memory, a processor, and an emergency report generation program stored in the memory and executable on the processor, wherein the emergency report generation program is configured to implement the steps of the emergency report generation method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores an emergency report generation program, and when the emergency report generation program is executed by a processor, the steps of the emergency report generation method according to any one of claims 1 to 6 are implemented.

Images