Emergency disposal decision information generation method and device, equipment and storage medium

Abstract

The invention provides an emergency disposal decision information generation method and device, equipment and a storage medium, and the method comprises the steps: obtaining historical emergency disposal cases, and generating an emergency disposal case corpus based on the historical emergency disposal cases; constructing a model training set based on the emergency disposal case corpus, and performing model training by using the model training set to obtain an auxiliary decision model; performing entity relationship extraction on pre-configured emergency plan information, and constructing an emergency disposal knowledge graph based on the entity relationship; and according to event description information of a to-be-processed event, emergency disposal decision information of the to-be-processed event is determined based on the auxiliary decision model and the emergency disposal knowledge graph. By adopting the method, the effectiveness and the accuracy of the emergency disposal decision information are improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method, device, equipment and storage medium for generating emergency response decision information. Background Art

[0002] With the expansion of the Internet and the increase in the complexity of enterprise management, various types of network emergencies and emergency risks are also increasing. Enterprises urgently need to establish an efficient emergency response mechanism to quickly identify, evaluate, make decisions and respond to emergencies through scientific methods and technical means to ensure the continuity of system services and reduce various losses.

[0003] The prior art solution closest to the technical solution of this application is a case matching-based network security operation decision-making assistance method (application publication number: CN 117350288A) disclosed in January 2024. This method first constructs a decision case training corpus and uses the word2vec model for training, then calculates the similarity between the event description of the target event and all training corpora, and selects several decision case corpora with the highest similarity to provide auxiliary decision-making basis for emergency handling of the event.

[0004] However, in the above scheme, only relatively simple text preprocessing is performed on the acquired decision case data, and model training is performed based on the simply processed data, which will cause the model training to only stay on the surface of the data and fail to reflect the characteristic attributes of the case data, resulting in the trained model being unable to accurately realize data prediction, thereby reducing the effectiveness of the model; secondly, the word2vec used in the prior art is a basic machine learning model, which ignores contextual information and can only process each word independently, and cannot represent the complex mapping relationship between emergencies and handling decisions, which will also make the model unable to accurately realize data prediction, further resulting in a reduction in the effectiveness and accuracy of the emergency handling decision information generated by the above model. Summary of the invention

[0005] In view of this, an object of the present invention is to provide a method, device, equipment and storage medium for generating emergency response decision information, so as to improve the effectiveness and accuracy of emergency response decision information.

[0006] In a first aspect, an embodiment of the present application provides a method for generating emergency response decision information, the method comprising:

[0007] Acquire historical emergency response cases, and generate an emergency response case corpus based on the historical emergency response cases;

[0008] Building a model training set based on the emergency response case corpus, and using the model training set to perform model training to obtain an auxiliary decision model;

[0009] Extract entity relationships from pre-configured emergency plan information and build an emergency response knowledge graph based on the entity relationships;

[0010] According to the event description information of the event to be processed, the emergency handling decision information of the event to be processed is determined based on the auxiliary decision model and the emergency handling knowledge graph.

[0011] Optionally, generating an emergency response case corpus based on the historical emergency response cases includes:

[0012] Preprocessing the historical emergency response cases to obtain optimized emergency response cases;

[0013] The text sequence of the optimized emergency response case is extracted, and word segmentation and label mapping processing are performed on the text sequence of the optimized emergency response case to generate the emergency response case corpus.

[0014] Optionally, constructing a model training set based on the emergency response case corpus, and using the model training set to perform model training to obtain an auxiliary decision model includes:

[0015] Get a pre-trained model to initialize model parameters;

[0016] Using all the corpus of the emergency response case corpus as a model training set to train the pre-trained model and optimize the model parameters;

[0017] The pre-trained model after training is saved as the auxiliary decision model.

[0018] Optionally, before extracting entity relationships from pre-configured emergency plan information, the method further includes:

[0019] The emergency response plan information is formulated based on experts' prior handling experience and basic theoretical knowledge in the field of emergency response.

[0020] Optionally, extracting entity relationships from pre-configured emergency plan information includes:

[0021] Preprocessing the emergency plan information to obtain optimized plan information;

[0022] Extract key entities in the optimization plan information by using named entity recognition technology;

[0023] Extract entity relationships between key entities through rule-based relationship extraction technology;

[0024] Post-process the entity relationships between key entities.

[0025] Optionally, constructing an emergency response knowledge graph based on the entity relationship includes:

[0026] Construct fault information table and risk scheme mapping table;

[0027] Based on the entity relationship between the key entities, each field in the fault information table and the risk plan mapping table is filled to form the emergency response knowledge graph.

[0028] Optionally, the determining, according to the event description information of the event to be processed, the emergency handling decision information of the event to be processed based on the auxiliary decision model and the emergency handling knowledge graph includes:

[0029] Preprocessing the event description information of the event to be processed to extract the text sequence of the event to be processed;

[0030] According to the sequence of texts of the event to be processed, determining first emergency handling information based on the auxiliary decision model;

[0031] According to the to-be-processed event text sequence, second emergency handling information is determined based on the emergency handling knowledge graph;

[0032] The first emergency handling information and the second emergency handling information are integrated to obtain emergency handling decision information of the event to be handled.

[0033] In a second aspect, an embodiment of the present application provides an emergency response decision information generation device, the device comprising:

[0034] A corpus generation module, used to obtain historical emergency response cases and generate an emergency response case corpus based on the historical emergency response cases;

[0035] A model training module, used to construct a model training set based on the emergency response case corpus, and use the model training set to perform model training to obtain an auxiliary decision model;

[0036] A knowledge graph construction module is used to extract entity relationships from pre-configured emergency plan information and construct an emergency response knowledge graph based on the entity relationships;

[0037] A decision information generation module is used to determine the emergency response decision information of the event to be processed based on the event description information of the event to be processed, the auxiliary decision model and the emergency response knowledge graph.

[0038] Optionally, generating an emergency response case corpus based on the historical emergency response cases includes:

[0039] Preprocessing the historical emergency response cases to obtain optimized emergency response cases;

[0040] The text sequence of the optimized emergency response case is extracted, and word segmentation and label mapping processing are performed on the text sequence of the optimized emergency response case to generate the emergency response case corpus.

[0041] Optionally, constructing a model training set based on the emergency response case corpus, and using the model training set to perform model training to obtain an auxiliary decision model includes:

[0042] Get a pre-trained model to initialize model parameters;

[0043] Using all the corpus of the emergency response case corpus as a model training set to train the pre-trained model and optimize the model parameters;

[0044] The pre-trained model after training is saved as the auxiliary decision model.

[0045] Optionally, before extracting entity relationships from pre-configured emergency plan information, the method further includes:

[0046] The emergency response plan information is formulated based on experts' prior handling experience and basic theoretical knowledge in the field of emergency response.

[0047] Optionally, extracting entity relationships from pre-configured emergency plan information includes:

[0048] Preprocessing the emergency plan information to obtain optimized plan information;

[0049] Extract key entities in the optimization plan information by using named entity recognition technology;

[0050] Extract entity relationships between key entities through rule-based relationship extraction technology;

[0051] Post-process the entity relationships between key entities.

[0052] Optionally, constructing an emergency response knowledge graph based on the entity relationship includes:

[0053] Construct fault information table and risk scheme mapping table;

[0054] Based on the entity relationship between the key entities, each field in the fault information table and the risk plan mapping table is filled to form the emergency response knowledge graph.

[0055] Optionally, the determining, according to the event description information of the event to be processed, the emergency handling decision information of the event to be processed based on the auxiliary decision model and the emergency handling knowledge graph includes:

[0056] Preprocessing the event description information of the event to be processed to extract the text sequence of the event to be processed;

[0057] According to the sequence of texts of the event to be processed, determining first emergency handling information based on the auxiliary decision model;

[0058] According to the to-be-processed event text sequence, second emergency handling information is determined based on the emergency handling knowledge graph;

[0059] The first emergency handling information and the second emergency handling information are integrated to obtain emergency handling decision information of the event to be handled.

[0060] In the third aspect, an embodiment of the present application provides a computer device, comprising: a processor, a memory and a bus, the memory storing machine-readable instructions executable by the processor, when the computer device is running, the processor and the memory communicate through the bus, and when the machine-readable instructions are executed by the processor, the steps of the emergency response decision information generation method described in any optional implementation manner of the first aspect above are performed.

[0061] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the emergency response decision information generation method described in any optional implementation manner of the first aspect above are executed.

[0062] The technical solution provided by this application includes but is not limited to the following beneficial effects:

[0063] First, this application generates an emergency response case corpus through the emergency response cases accumulated by the enterprise over a long period of time, and completes the training of the auxiliary decision-making model. Then, an emergency plan is formulated based on the emergency response experience and theoretical knowledge, and an emergency response knowledge graph is constructed. Finally, based on the reasoning results of the auxiliary decision-making model and the retrieval results of the emergency response knowledge graph, the intelligently recommended emergency response auxiliary decision-making information for the target event is integrated and matched to ensure the effectiveness and accuracy of the emergency response decision-making information, and further help the disposal personnel to take effective and controllable disposal measures in a timely manner, quickly control the scope of the incident impact, and ensure the continuity of system services.

[0064] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0066] Figure 1 A flowchart of a method for generating emergency response decision information provided by Embodiment 1 of the present invention is shown;

[0067] Figure 2 A flowchart of a method for generating an emergency response case corpus provided in the first embodiment of the present invention is shown;

[0068] Figure 3 A flowchart of a decision-making support model training method provided by the first embodiment of the present invention is shown;

[0069] Figure 4 A flow chart of an entity relationship extraction method provided by the first embodiment of the present invention is shown;

[0070] Figure 5 A flowchart of a method for constructing an emergency response knowledge graph provided by the first embodiment of the present invention is shown;

[0071] Figure 6 A flowchart of a method for determining emergency response decision information provided by Embodiment 1 of the present invention is shown;

[0072] Figure 7 A schematic diagram showing the structure of an emergency response decision information generating device provided by a second embodiment of the present invention is shown;

[0073] Figure 8 A schematic diagram of the structure of a computer device provided in Embodiment 3 of the present invention is shown. DETAILED DESCRIPTION

[0074] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present invention.

[0075] Embodiment 1

[0076] In order to solve the problems of high time urgency, limited expert resources, and inefficient implementation of internal emergency response experience in emergency response work, natural language processing technology, large model technology, and knowledge graph technology are introduced to provide a method for generating emergency response decision information by generating an emergency response case corpus, training an auxiliary decision model, building an emergency response knowledge graph, and matching emergency response auxiliary decisions. Figure 1 The flowchart of a method for generating emergency response decision information provided by the first embodiment of the present invention is shown to describe the contents of the first embodiment of the present application in detail.

[0077] See also Figure 1 As shown, Figure 1 A flowchart of a method for generating emergency response decision information provided by the first embodiment of the present invention is shown, wherein the method comprises steps S101 to S104:

[0078] S101: Acquire historical emergency response cases, and generate an emergency response case corpus based on the historical emergency response cases.

[0079] Specifically, we collect historical emergency response cases accumulated over a long period of time, pre-process the text of historical emergency response cases, extract case description text sequences and response decision text sequences, and perform word segmentation and label mapping on the response decision text sequences to generate an emergency response case corpus. New practical cases need to be processed in a timely manner and entered into the corpus.

[0080] Text preprocessing includes text cleaning, standardization, grammar correction and segmentation. Case description includes name, fault type and fault phenomenon information. Disposal decision includes personnel, role, risk level, fault level and disposal measure information.

[0081] The emergency response case corpus is used to centrally store processed emergency response case language materials and provide data for the training process of the decision support model. This application is based on the emergency response cases accumulated in the work practice within the enterprise. Through text preprocessing, the case description text sequence and the disposal decision text sequence are divided, and the word segmentation and label mapping method is used to convert the disposal decision text sequence into a label vector. Then the case description text sequence is used as the input of the model, and the disposal decision label vector is used as the expected output of the model. The two together constitute the emergency response case training corpus, which is used to train the decision support model so that the model can correctly feedback the disposal decision attribute information based on the input case text description. The emergency response decision case corpus is generated through three steps: text preprocessing, word segmentation and label mapping, and corpus generation.

[0082] S102: constructing a model training set based on the emergency response case corpus, and performing model training using the model training set to obtain an auxiliary decision model.

[0083] Specifically, we use the emergency response case corpus to build a training set, train the ALBERT natural language processing model to adapt to the decision-making support classification task, and obtain the decision-making support model. Through periodic training and updating of model parameters, we ensure that the model can use the new data in the corpus to optimize the classification performance.

[0084] S103: Extract entity relationships from pre-configured emergency plan information, and construct an emergency response knowledge graph based on the entity relationships.

[0085] Specifically, an emergency plan is a work plan prepared in advance for scientific, orderly and rapid response to emergencies, which includes text descriptions such as fault type, fault phenomenon, risk analysis and disposal plan. The emergency response knowledge graph is used to integrate the data of emergency response expert experience and technical theoretical knowledge, retrieve risk analysis, disposal plan and other disposal decision information based on the fault type and fault phenomenon of the emergency, and provide standardized and unified emergency response decision standards for different types of emergencies.

[0086] First, based on the long-term accumulation of expert prior experience and basic theoretical knowledge in the field of emergency response that can be effectively organized and reused, a pre-trained plan-making model is used to formulate emergency plans covering multiple information systems and multiple risk failure scenarios. Then, a pipeline-based entity extraction method is used to automatically extract entity relationships between emergency plans and mine triples and relationship types. On this basis, a database is designed based on entity relationships, and the OceanBase database management system is used to store and manage emergency storage knowledge graph data.

[0087] Through the three steps of emergency plan formulation, entity relationship extraction, and database design, the emergency response knowledge graph is constructed to achieve effective integration and application of emergency response knowledge. Emergency plans are formulated based on emergency response experience and theoretical knowledge, and unified and standardized emergency response process standards are provided for various emergencies. The entity relationship extraction method and relational database design method are used to construct the emergency response knowledge graph, realizing the automatic extraction of entity relationships between emergency plans and the centralized storage of entity data.

[0088] This application uses HanLP word segmentation technology to generate decision-making case training corpus, uses the ALBERT natural language processing large model to build an auxiliary decision model, and uses a pipeline-based entity relationship extraction method to build an emergency response knowledge graph. Among them, HanLP is a high-performance, easily scalable natural language processing toolkit, mainly used for Chinese word segmentation tasks; ALBERT is an efficient, lightweight pre-trained language model, and the BERT-based Transformer architecture achieves small memory usage and high performance. The model can map case description text data layer by layer to label vectors that refer to decision attribute information to characterize the complex mapping relationship between emergencies and handling decisions, thereby realizing the function of inferring handling decisions based on event descriptions; the pipeline-based entity relationship extraction method is a common method in the process of building a knowledge graph. It aims to extract entity relationships from the unstructured text data of emergency plans to build an emergency response knowledge graph.

[0089] S104: According to the event description information of the event to be processed, the emergency handling decision information of the event to be processed is determined based on the auxiliary decision model and the emergency handling knowledge graph.

[0090] Specifically, emergency response decision support matching aims to match the intelligently recommended emergency response decision information for the event based on the target event description, the trained decision support model and the constructed emergency response knowledge graph, and provide this auxiliary information to decision makers to assist them in determining the response measures to be taken.

[0091] The text description information of the target event is obtained, and after the same text preprocessing as step S101, it is input into the auxiliary decision-making model trained in step S102, and the disposal label information with the highest classification probability is output. At the same time, the corresponding emergency plan data is retrieved through the emergency disposal knowledge graph constructed in step S103 as supplementary information. Then, through the effective integration of the two parts of information in step S104, the emergency disposal auxiliary decision-making information intelligently matched to the target event is formed.

[0092] Compared with the existing technical solutions that use traditional machine learning algorithms to calculate text similarity to simply mine the data relationship between event descriptions and auxiliary decision-making, this application uses advanced AI large model technology to train the model and intelligently mine higher-dimensional and effective feature representation relationships between data. At the same time, this application constructs an emergency response knowledge graph based on the formulated emergency plan to provide useful reference information for emergency response decisions, and supplements the auxiliary decision-making output by the model to provide accurate, detailed and reasonable emergency response measures. The technical solution of this application helps to fully reuse the prior experience of emergency response accumulated by the enterprise over a long period of time, so as to intelligently recommend emergency response decision-making information for specific emergencies, and respond quickly and deal with them effectively.

[0093] In an alternative embodiment, see Figure 2 As shown, Figure 2 A flowchart of a method for generating an emergency response case corpus provided by the first embodiment of the present invention is shown, wherein the method for generating an emergency response case corpus based on the historical emergency response cases comprises steps S201 to S202:

[0094] S201: Pre-process the historical emergency response cases to obtain optimized emergency response cases.

[0095] Specifically, the optimization of emergency response cases includes case description text sequences and disposal decision text sequences. The text preprocessing of emergency response cases includes data cleaning, standardization, grammar correction and segmentation. First, the text data is cleaned to remove special characters, stop words and non-text elements. Secondly, standardization is performed to convert the text to lowercase, remove the difference in uppercase and lowercase letters, and use stem extraction and word form restoration technology to simplify the words to basic forms. Then, spelling and grammatical errors in the text are corrected to ensure the consistency of the expression. Finally, the corresponding segments are divided to clarify the name, fault type, fault phenomenon and other information to form a case description text sequence; the corresponding segments are divided to clarify the personnel, roles, responsibilities, risk level, fault level, disposal measures and other information to form a disposal decision text sequence.

[0096] S202: extracting the text sequence of the optimized emergency handling case, performing word segmentation and label mapping processing on the text sequence of the optimized emergency handling case, and generating the emergency handling case corpus.

[0097] Specifically, the disposal decision text sequence is processed by word segmentation and label mapping to generate an emergency disposal case corpus. The auxiliary decision model constructed by the present invention belongs to a text classification model. Therefore, before training the model, it is necessary to construct a classification label system to characterize the emergency disposal decision information inferred by the model. The disposal decision text sequence obtained after text preprocessing is taken as the label mapping object, and the HanLP technology is first used to perform word segmentation processing to generate a semantically independent keyword sequence. In order to improve the accuracy of word segmentation, the present invention organizes the field professional vocabulary involved in the emergency disposal process to form a dictionary for application in word segmentation operations. Then, the corresponding vocabulary is extracted from the keyword sequence to match the five characteristic attributes of disposal personnel, role, risk level, fault level, and disposal measures as the auxiliary decision information provided by the expected auxiliary decision model for the target event. The vector output by the model is a vector of dimension 1*5. The selection result of personnel is mapped to the first output value of the vector, representing the candidate personnel selected from m emergency response personnel, where m is the total number of emergency response personnel of the enterprise; the selection result of role is mapped to the second output value of the vector, representing the role selected from n emergency response roles, where n is the total number of emergency response role types configured by the enterprise; the selection result of risk level is mapped to the third output value of the vector, with value 0 representing "high", value 1 representing "medium", and value 2 representing "low"; the selection result of fault level is mapped to the fourth output value of the vector, with value 0 representing "level 1", value 1 representing "level 2", value 2 representing "level 3", and value 3 representing "level 4"; the selection result of disposal measures is mapped to the fifth output value of the vector, representing the disposal measures selected from k emergency disposal measures, where k is the total number of disposal measures taken in historical emergency disposal cases. According to the above steps, the disposal decision text sequence is mapped and converted into a disposal decision label vector.

[0098] The case description text sequence is entered as the input of the auxiliary decision model; at the same time, the disposal decision label vector is entered as the output of the auxiliary decision model. The two together constitute the decision case training corpus, which is centrally stored in the emergency disposal case corpus. It should be noted that the corpus construction of this application is a continuous updating and iterative process. It is necessary to timely pre-process the text, segment and label map the newly occurring case data into the corpus based on the disposal experience accumulated continuously in practice.

[0099] In an alternative embodiment, see Figure 3 As shown, Figure 3 A flowchart of a decision support model training method provided in the first embodiment of the present invention is shown, wherein the model training set is constructed based on the emergency response case corpus, and the decision support model is obtained by model training using the model training set, including steps S301 to S303:

[0100] S301: Obtain a pre-trained model to initialize model parameters.

[0101] Specifically, the decision-making support model is used to intelligently infer some important emergency response support decision-making information for the target event to improve the response speed and accuracy of the emergency response. This application uses the ALBERT model (a lightweight variant based on BERT) to train the decision-making support model. Download the official pre-trained model of ALBERT from the Github repository of Google Research to initialize the model parameters, and use the entire corpus of the emergency response case corpus as the model training set.

[0102] S302: Using the entire corpus of the emergency response case corpus as a model training set to train the pre-trained model and optimize the model parameters.

[0103] Specifically, during the training process, the following key parameters need to be set:

[0104] The vocab_size parameter is used to define the vocabulary size of the model, that is, the maximum number of different tokens that the model can recognize, which is directly related to the model's ability to understand and generate text. The training set constructed in this application is small in scale and has a high concentration of domain terms, so the selected vocab_size can be set smaller accordingly to reduce the scale of model parameters and reduce the risk of overfitting. This application uses cross-validation technology to evaluate the model performance when vocab_size ranges from 10,000 to 30,000, and sets vocab_size = 10,000.

[0105] The embedding_size parameter is used to define the dimension of the model word embedding, that is, the dimension of the embedding vector that each token will be converted into. During the training process, these vectors will learn rich language features. The training set constructed by this application is small in scale and does not show overly complex language features. Therefore, this application sets embedding_size = 144 to reduce the scale of model parameters and reduce the risk of overfitting.

[0106] The hidden_size parameter is used to define the size of the model encoder layer and pooling layer, that is, the length of each layer of the network vector inside the model, which directly affects the model's ability to process information. For text classification tasks, a too large hidden_size will increase the complexity and computational cost of the model, resulting in unnecessary resource consumption. Based on a comprehensive consideration of model performance, task requirements, and computing resources, set hidden_size to 144.

[0107] The num_hidden_layers parameter is used to define the number of hidden layers in the model encoder, which determines how many consecutive hidden layers the input information will be processed through. Generally, adding hidden layers allows the model to learn richer feature levels, but it also leads to an increase in model complexity and computational cost. This application uses cross-validation technology to evaluate the model performance when num_hidden_layers ranges from 1 to 6, setting num_hidden_layers = 3.

[0108] The num_attention_heads parameter is used to define the number of heads in each attention layer in the multi-head attention mechanism. This mechanism allows the model to simultaneously focus on multiple subsets of information at different positions in the input sequence, thereby enhancing its understanding ability. This paper follows the common configuration of the BERT model using 12 attention heads and sets num_attention_heads = 12.

[0109] The num_workers parameter is used to define the number of parallel threads during training, and is set based on the available training resource performance. Considering the computing resource limitations, set num_workers = 8.

[0110] The patience parameter is used to define how many consecutive epochs the model will stop training when the dataset loss no longer decreases when using the early stopping method. In the case of balancing preventing overfitting and avoiding underfitting, set patience = 20.

[0111] S303: Save the pre-trained model after training as the auxiliary decision model.

[0112] Specifically, after completing the model training, the model is saved in the ckpt file format. In addition, since the emergency response case corpus continues to input new case corpora, periodic training and updating of existing models can effectively utilize the new data to optimize model performance.

[0113] In an optional implementation, before extracting entity relationships from pre-configured emergency plan information, the method further includes:

[0114] The emergency response plan information is formulated based on experts' prior handling experience and basic theoretical knowledge in the field of emergency response.

[0115] Specifically, the present invention combines the emergency response experience accumulated through long-term practice within the enterprise and the technical theoretical knowledge in the field, extracts and integrates effective information, and formulates an emergency plan. The emergency plan lists the risk information and possible emergency response plans that may be caused by various common emergencies in the form of text description.

[0116] For example, for an event that belongs to the product failure type, the failure phenomenon is "MP call interface failure". For this event, the risk analysis information recorded in the emergency plan is "upstream service problems, automated operation and maintenance system is unavailable", and the deployed emergency response plan is "check the call chain, check the abnormal points, confirm the cause of the abnormality; if necessary, urgently contact the product team of the automated operation and maintenance system to check the corresponding problems and follow up on the resolution of the problems".

[0117] By deploying emergency response decisions in advance, the emergency plan can help to provide timely feedback on effective and controllable response measures after an emergency occurs, further improve the response speed and accuracy of emergency response, and ensure the continuity of system services.

[0118] In an alternative embodiment, see Figure 4 As shown, Figure 4 A flowchart of an entity relationship extraction method provided by the first embodiment of the present invention is shown, wherein the entity relationship extraction of pre-configured emergency plan information includes steps S401 to S404:

[0119] S401: Preprocess the emergency plan information to obtain optimized plan information.

[0120] Specifically, the emergency plan text is uniformly preprocessed, including sentence segmentation, word segmentation and part-of-speech tagging.

[0121] S402: Extract key entities in the optimization plan information through named entity recognition technology.

[0122] Specifically, named entity recognition technology is used to extract key entities in the emergency plan, including fault type, fault phenomenon, risk analysis and disposal plan. Some entity data in the emergency plan are shown in the following table:

[0123]

[0124] S404: Extract entity relationships between key entities through rule-based relationship extraction technology.

[0125] Specifically, rule-based relation extraction technology is used to extract relations between entities.

[0126] S404: Post-process the entity relationships between key entities.

[0127] Specifically, the extraction results are post-processed, including entity deduplication, relationship merging and relationship reasoning, to improve the accuracy and completeness of the extraction results. The entity relationship triples of the emergency plan extracted according to the above steps are shown in the following table:

[0128]

[0129] In an alternative embodiment, see Figure 5 As shown, Figure 5 A flowchart of a method for constructing an emergency response knowledge graph provided by the first embodiment of the present invention is shown, wherein the method for constructing an emergency response knowledge graph based on the entity relationship includes steps S501 to S502:

[0130] S501: Construct a fault information table and a risk solution mapping table.

[0131] Specifically, the construction of the emergency response knowledge graph of this application relies on a relational database, so the OceanBase database management system is chosen for data storage and data management. The OceanBase database is a high-performance, highly available distributed relational database that can store data in different data tables to achieve flexibility in accessing and modifying data.

[0132] According to the extracted emergency plan entity relationship, this application designs a fault information table and a risk-solution mapping table. The fault information table uses the fault phenomenon number as the primary key and uses the risk-solution mapping number as the foreign key to realize the association between emergency plan entities.

[0133] S502: Fill in the fields in the fault information table and the risk solution mapping table based on the entity relationships between the key entities to form the emergency response knowledge graph.

[0134] Specifically, in the attribute design of the fault information table, this table contains four fields: fault phenomenon number, fault phenomenon description, risk-solution mapping number, and fault type description, which are used to store the specific information of the event fault. The risk-solution mapping table includes three fields: risk-solution mapping number, risk analysis description, and disposal solution description, which are used to store the corresponding emergency disposal knowledge information. The fault information table is shown in the following table:

[0135]

[0136] The risk-solution mapping table is shown in the following table

[0137]

[0138] The emergency response knowledge graph constructed in this application brings together a wide range of emergency response expert experience and field theoretical knowledge. By extracting fault type and fault phenomenon information from the case description text sequence of the target event, the knowledge graph can retrieve the corresponding risk analysis and response plan. This step aims to achieve the unification and standardization of emergency response process standards and improve the accuracy of emergency response.

[0139] In an alternative embodiment, see Figure 6 As shown, Figure 6 A flowchart of a method for determining emergency response decision information provided by the first embodiment of the present invention is shown, wherein the emergency response decision information of the event to be processed is determined based on the auxiliary decision model and the emergency response knowledge graph according to the event description information of the event to be processed, including steps S601 to S604:

[0140] S601: Preprocessing the event description information of the event to be processed to extract the text sequence of the event to be processed.

[0141] Specifically, based on the text preprocessing in the corpus generation stage, the text description of the target event is cleaned, standardized, grammatically corrected, and segmented to clarify information such as the name, fault type, and fault phenomenon, forming a case description text sequence of the event, that is, the text sequence of the event to be processed.

[0142] S602: Determine first emergency handling information based on the auxiliary decision model according to the text sequence of the event to be processed.

[0143] Specifically, the case description text sequence in the target event corpus is input into the trained auxiliary decision-making model, and the model outputs the corresponding classification label vector, which feeds back the five types of emergency response attribute information inferred by the model, namely, the first emergency response information, including personnel, role, risk level, fault level, and handling measures.

[0144] S603: Determine second emergency response information based on the emergency response knowledge graph according to the event text sequence to be processed.

[0145] Specifically, extract the fault phenomenon description in the target event corpus, query all the fault phenomenon description fields stored in the knowledge graph, calculate the similarity with the target event fault phenomenon description, select the fault phenomenon description with the highest similarity, and map it to the corresponding fault phenomenon sequence number field value. Use the fault phenomenon sequence number as the primary key to query the risk-solution mapping sequence number in the fault information table; then use the queried risk-solution mapping sequence number as the primary key to query the risk analysis description information and disposal solution description information stored in the risk-solution mapping table, that is, the second emergency disposal information.

[0146] S604: Integrate the first emergency handling information and the second emergency handling information to obtain emergency handling decision information of the event to be handled.

[0147] Specifically, the emergency response decision information inferred by the auxiliary decision-making model (including five types of information, namely personnel, role, risk level, fault level, and disposal measures) and the emergency response decision information retrieved from the emergency response knowledge graph (including two types of information, namely risk analysis and disposal plan) are summarized and integrated to form the emergency response decision information recommended for the target event, which is used to assist decision makers in making disposal decisions and improve the response speed and accuracy of emergency response.

[0148] Embodiment 2

[0149] See also Figure 7 As shown, Figure 7 The structure diagram of an emergency response decision information generating device provided by the second embodiment of the present invention is shown, wherein the device comprises:

[0150] A corpus generation module 701 is used to obtain historical emergency response cases and generate an emergency response case corpus based on the historical emergency response cases;

[0151] A model training module 702 is used to construct a model training set based on the emergency response case corpus, and use the model training set to perform model training to obtain an auxiliary decision model;

[0152] The knowledge graph construction module 703 is used to extract entity relationships from pre-configured emergency plan information and construct an emergency response knowledge graph based on the entity relationships;

[0153] The decision information generation module 704 is used to determine the emergency response decision information of the event to be processed based on the event description information of the event to be processed, the auxiliary decision model and the emergency response knowledge graph.

[0154] In an optional implementation, generating an emergency response case corpus based on the historical emergency response cases includes:

[0155] Preprocessing the historical emergency response cases to obtain optimized emergency response cases;

[0156] The text sequence of the optimized emergency response case is extracted, and word segmentation and label mapping processing are performed on the text sequence of the optimized emergency response case to generate the emergency response case corpus.

[0157] In an optional implementation scheme, the step of constructing a model training set based on the emergency response case corpus and using the model training set to perform model training to obtain an auxiliary decision model includes:

[0158] Get a pre-trained model to initialize model parameters;

[0159] Using all the corpus of the emergency response case corpus as a model training set to train the pre-trained model and optimize the model parameters;

[0160] The pre-trained model after training is saved as the auxiliary decision model.

[0161] In an optional implementation, before extracting entity relationships from pre-configured emergency plan information, the method further includes:

[0162] The emergency response plan information is formulated based on experts' prior handling experience and basic theoretical knowledge in the field of emergency response.

[0163] In an optional implementation scheme, extracting entity relationships from pre-configured emergency plan information includes:

[0164] Preprocessing the emergency plan information to obtain optimized plan information;

[0165] Extracting key entities in the optimization plan information by using named entity recognition technology;

[0166] Extract entity relationships between key entities through rule-based relationship extraction technology;

[0167] Post-process the entity relationships between key entities.

[0168] In an optional implementation scheme, constructing an emergency response knowledge graph based on the entity relationship includes:

[0169] Construct fault information table and risk scheme mapping table;

[0170] Based on the entity relationship between the key entities, each field in the fault information table and the risk plan mapping table is filled to form the emergency response knowledge graph.

[0171] In an optional implementation scheme, the step of determining the emergency handling decision information of the event to be handled based on the event description information of the event to be handled and the auxiliary decision model and the emergency handling knowledge graph includes:

[0172] Preprocessing the event description information of the event to be processed to extract the text sequence of the event to be processed;

[0173] According to the sequence of texts of the event to be processed, determining first emergency handling information based on the auxiliary decision model;

[0174] According to the to-be-processed event text sequence, second emergency handling information is determined based on the emergency handling knowledge graph;

[0175] The first emergency handling information and the second emergency handling information are integrated to obtain emergency handling decision information of the event to be handled.

[0176] Embodiment 3

[0177] Based on the same application concept, see Figure 8 As shown, Figure 8 FIG. 4 shows a schematic diagram of the structure of a computer device provided by Embodiment 3 of the present invention, wherein: Figure 8 As shown, a computer device 800 provided in Embodiment 3 of the present application includes:

[0178] A processor 801, a memory 802 and a bus 803, wherein the memory 802 stores machine-readable instructions executable by the processor 801. When the computer device 800 is running, the processor 801 communicates with the memory 802 via the bus 803. When the processor 801 is running, the machine-readable instructions execute the steps of the method for generating emergency response decision information shown in the first embodiment.

[0179] Embodiment 4

[0180] Based on the same application concept, an embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the emergency response decision information generation method described in any one of the above embodiments are executed.

[0181] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0182] The computer program product for generating emergency response decision information provided in an embodiment of the present invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the method described in the previous method embodiment. The specific implementation can be found in the method embodiment, which will not be repeated here.

[0183] The emergency response decision information generation device provided in the embodiment of the present invention can be specific hardware on the device or software or firmware installed on the device. The implementation principle and technical effects of the device provided in the embodiment of the present invention are the same as those of the aforementioned method embodiment. For the sake of brief description, for matters not mentioned in the device embodiment, reference can be made to the corresponding contents in the aforementioned method embodiment. Technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices and units described above can all refer to the corresponding processes in the aforementioned method embodiment, and will not be repeated here.

[0184] In the embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0185] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0186] In addition, each functional unit in the embodiment provided by the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0187] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0188] It should be noted that similar numbers and letters represent similar items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are only used to distinguish the description and are not to be understood as indicating or implying relative importance.

[0189] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, those skilled in the art should understand that any person skilled in the art can still modify the technical solutions described in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or perform equivalent replacements on some of the technical features thereof; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. They should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for generating emergency response decision information, characterized in that: The method comprises: Acquire historical emergency response cases, and generate an emergency response case corpus based on the historical emergency response cases; Building a model training set based on the emergency response case corpus, and using the model training set to perform model training to obtain an auxiliary decision model; Extract entity relationships from pre-configured emergency plan information and build an emergency response knowledge graph based on the entity relationships; According to the event description information of the event to be processed, the emergency handling decision information of the event to be processed is determined based on the auxiliary decision model and the emergency handling knowledge graph.

2. The method according to claim 1, characterized in that The generating of the emergency response case corpus based on the historical emergency response cases includes: Preprocessing the historical emergency response cases to obtain optimized emergency response cases; The text sequence of the optimized emergency response case is extracted, and word segmentation and label mapping processing are performed on the text sequence of the optimized emergency response case to generate the emergency response case corpus.

3. The method according to claim 1, characterized in that: The step of constructing a model training set based on the emergency response case corpus and using the model training set to perform model training to obtain an auxiliary decision model includes: Get a pre-trained model to initialize model parameters; Using all the corpus of the emergency response case corpus as a model training set to train the pre-trained model and optimize the model parameters; The pre-trained model after training is saved as the auxiliary decision model.

4. The method according to claim 1, characterized in that: Before extracting entity relationships from the pre-configured emergency plan information, the method further includes: The emergency response plan information is formulated based on experts' prior handling experience and basic theoretical knowledge in the field of emergency response.

5. The method according to claim 1, characterized in that The extracting of entity relationships from pre-configured emergency plan information includes: Preprocessing the emergency plan information to obtain optimized plan information; Extracting key entities in the optimization plan information by using named entity recognition technology; Extract entity relationships between key entities through rule-based relationship extraction technology; Post-process the entity relationships between key entities.

6. The method according to claim 1, characterized in that The constructing of the emergency response knowledge graph based on the entity relationship includes: Construct fault information table and risk scheme mapping table; Based on the entity relationship between the key entities, each field in the fault information table and the risk plan mapping table is filled to form the emergency response knowledge graph.

7. The method according to claim 1, characterized in that The step of determining the emergency handling decision information of the event to be handled based on the event description information of the event to be handled and the auxiliary decision model and the emergency handling knowledge graph includes: Preprocessing the event description information of the event to be processed to extract the text sequence of the event to be processed; According to the sequence of texts of the event to be processed, determining first emergency handling information based on the auxiliary decision model; According to the to-be-processed event text sequence, second emergency handling information is determined based on the emergency handling knowledge graph; The first emergency handling information and the second emergency handling information are integrated to obtain emergency handling decision information of the event to be handled.

8. An emergency response decision information generation device, characterized in that: The device comprises: A corpus generation module, used to obtain historical emergency response cases and generate an emergency response case corpus based on the historical emergency response cases; A model training module, used to construct a model training set based on the emergency response case corpus, and use the model training set to perform model training to obtain an auxiliary decision model; A knowledge graph construction module is used to extract entity relationships from pre-configured emergency plan information and construct an emergency response knowledge graph based on the entity relationships; A decision information generation module is used to determine the emergency response decision information of the event to be processed based on the event description information of the event to be processed, the auxiliary decision model and the emergency response knowledge graph.

9. A computer device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the method for generating emergency response decision information as described in any one of claims 1 to 7 are performed.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, executes the steps of the method for generating emergency response decision information as described in any one of claims 1 to 7.

Images