Geological disaster emergency response scheme generation method based on domain knowledge graph

Abstract

The invention is suitable for the technical field of geological disasters, and provides a geological disaster emergency response scheme generation method based on a domain knowledge graph, and the method comprises the steps: determining an optimal path set of each entity of a target geological disaster event in a geological disaster emergency response knowledge graph, and determining the weight of each entity; screening out a plurality of reference geological disaster events of the target geological disaster event from historical geological disaster events used when the geological disaster emergency response knowledge map is constructed based on weights by taking geological disaster mechanism and effect similarity as constraints; screening out a matched geological disaster event of the target geological disaster event from the plurality of reference geological disaster events based on the graph similarity; and generating an emergency response scheme of the target geological disaster event based on the geological disaster emergency response knowledge map and the matched geological disaster event. According to the invention, the matching precision of the geological disaster event can be improved, and the emergency response scheme with high feasibility can be automatically generated based on the matching result.

Description

Technical Field

[0001] This application belongs to the technical field of geological disasters, and particularly relates to a method for generating a geological disaster emergency response plan based on a domain knowledge graph. Background Technique

[0002] China is located in a complex and changeable geological environment, with frequent geological disasters and strong destructiveness, posing a serious threat to people's lives and property safety and natural resources. In the whole process of geological disaster management, "monitoring → identification → early warning → emergency response", as the last line of defense for disaster management, the emergency response is directly related to life safety and social stability. Specifically, after a major geological disaster occurs, multiple parties immediately respond, conduct disaster situation statistics and reporting, the command headquarters formulates an emergency response plan based on the actual disaster situation, organizes multi-departmental coordinated rescue, evacuates the people within the scope affected by the geological disaster in a timely manner, conducts search and rescue work on the trapped public, and repairs the damaged important public facilities such as roads and transportation pipelines to save people's lives and property safety at the fastest speed. In this process, quickly generating an emergency response plan for the actual disaster situation is of great significance for the efficient progress of disaster response and rescue work.

[0003] At present, the methods for generating geological disaster emergency response plans are mainly divided into three categories: empirical discrimination method, numerical simulation method, and case-based reasoning method. The empirical discrimination method relies on professionals to conduct qualitative analysis in the face of actual situations. This method has extremely high requirements for the professional quality of personnel and is highly subjective. The numerical simulation method constructs a mathematical model based on the mechanism of geological disasters. Although it has a high level of refinement, it has high requirements for data quality and computing resources and is difficult to respond in real time. The case-based reasoning method generates an emergency response plan by algorithmically matching historical disaster cases, but it has high requirements for the accuracy of the matching algorithm.

[0004] With the development of natural language processing technology, many scholars have introduced technologies such as knowledge graphs and semantic similarity into the generation of emergency response plans for geological disaster emergency responses. Some scholars have introduced semantic similarity into case matching, which has broadened the matching dimension to a certain extent and mined some detailed information that cannot be captured by traditional matching methods. However, the essence of case matching based on semantic similarity is based on the similarity between words and cannot deeply understand the geological disaster mechanism in the text, resulting in low matching accuracy.

[0005] Some scholars have used knowledge graphs to manage the complex semantic knowledge in geological disaster emergency responses, constructed knowledge bases, and characterized the relevant attributes affecting emergency responses through entity reasoning. Then, they characterized the similarity of cases through the similarity of attributes, and finally matched the cases and inferred and generated emergency response plans. The attributes involved in geological disasters and their emergency response designs are extensive and complex, and there may be direct or potential relationships between some attributes. Attributes with different relationships have different impacts on geological disaster emergency responses, and different attributes with the same relationship also have different impacts on geological disaster emergency responses. However, existing methods lack consideration of the relationships between attributes and the changes in attribute weights in this relationship, resulting in low matching accuracy and poor implementability of the finally formulated emergency response plans. Summary of the Invention

[0006] The embodiments of the present application provide a method for generating a geological disaster emergency response plan based on a domain knowledge graph, which can solve the problems of low matching accuracy of geological disaster events and poor implementability of emergency response plans.

[0007] The embodiments of the present application provide a method for generating a geological disaster emergency response plan based on a domain knowledge graph, including:

[0008] For each entity of the target geological disaster event in the pre-constructed geological disaster emergency response knowledge graph, determine the optimal path set of the entity; the optimal path set includes the optimal path from each other entity in the geological disaster emergency response knowledge graph to the entity;

[0009] According to the optimal path set of each entity, determine the weight of each entity of the target geological disaster event in the geological disaster emergency response knowledge graph;

[0010] Taking the similarity of geological disaster mechanism actions as a constraint, based on the determined weights, screen out multiple reference geological disaster events of the target geological disaster event from the historical geological disaster events used when constructing the geological disaster emergency response knowledge graph; the entities of the geological disaster emergency response knowledge graph correspond to the attributes of the historical geological disaster events and the target geological disaster event;

[0011] Based on graph similarity, screen out the matching geological disaster events of the target geological disaster event from multiple reference geological disaster events;

[0012] Based on the geological disaster emergency response knowledge graph and the matching geological disaster events, generate an emergency response plan for the target geological disaster event.

[0013] Optionally, determining the optimal path set of the entity includes:

[0014] For each other entity in the geological disaster emergency response knowledge graph except for entities, determine the optimal path from the other entity to the entity through the path optimization calculation formula;

[0015] The path optimization calculation formula is:

[0016] PathBA = arg P∈Path_Set(B,A) max(F(P))

[0017] F(P) = IIS(P) / L(P)

[0018] IIS(P) = S(r B1 ) × S(r 12 ) × … S(r nA )

[0019] L(P) = |V| - 1

[0020] Among them, PathBA represents the optimal path from other entity B to entity A, Path_Set(B,A) represents the set of all paths from other entity B to entity A, IIS(P) represents the path semantic strength of path P in Path_Set(B,A), L(P) represents the path length of path P, S(r B1 ) represents the semantic strength of relationship r B1 in path P, S(r 12 ) represents the semantic strength of relationship r 12 in path P, S(r nA ) represents the semantic strength of relationship r nA in path P, r B1 represents the relationship between other entity B and entity e 1 , r 12 represents the relationship between entity e 1 and entity e 2 , r nA represents the relationship between entity e n and entity A, entity e 1 , e 2 and e n are entities on path P, |V| represents the number of entities included in path P.

[0021] Optionally, the geological disaster emergency response plan generation method further includes:

[0022] For entity A of the target geological disaster event and other entity B in the geological disaster emergency response knowledge graph, if there are both the optimal path PathAB from entity A to other entity B and the optimal path PathBA from other entity B to entity A, then the optimal path is retained through the following formula:

[0023]

[0024] Among them, Level(A) represents the level of entity A in the geological disaster emergency response knowledge graph, Level(B) represents the level of other entity B in the geological disaster emergency response knowledge graph, TC(PathAB) represents the path smoothness of PathAB, TC(PathBA) represents the path smoothness of PathBA, IIS(PathAB) represents the path semantic intensity of PathAB, exp(-L(PathAB)) represents the path length attenuation factor of PathAB, IIS(PathBA) represents the path semantic intensity of PathBA, and exp(-L(PathBA)) represents the path length attenuation factor of PathBA.

[0025] Optionally, according to the optimal path set of each entity, determine the weight of each entity of the target geological disaster event in the geological disaster emergency response knowledge graph, including:

[0026] For each entity of the target geological disaster event in the geological disaster emergency response knowledge graph, calculate the weight w of the entity through the following formula A :

[0027]

[0028] Among them, w 0 represents the initial weight, Path_Set(A) represents the optimal path set of entity A, e i represents the i-th entity on the optimal path Q, e i+1 represents the (i + 1)-th entity on the optimal path Q, r i(i+1) represents the relationship between e i and e i+1 , S(r i(i+1) ) represents the semantic intensity of r i(i+1) , represents the distance attenuation factor of the i-th entity e i on the optimal path Q, d(e i , A) represents the number of hops from the i-th entity e i on the optimal path Q to entity A.

[0029] Optionally, with the similarity of geological disaster mechanism effects as a constraint, based on the determined weights, screen out multiple reference geological disaster events of the target geological disaster event from the historical geological disaster events used when constructing the geological disaster emergency response knowledge graph, including:

[0030] For each historical geological disaster event used when constructing the geological disaster emergency response knowledge graph, calculate the mechanism similarity between the target geological disaster event and the historical geological disaster event through the mechanism similarity calculation formula;

[0031] Use the historical geological disaster events corresponding to the mechanism similarity with a numerical value greater than the preset similarity threshold as the reference geological disaster events for the target geological disaster event.

[0032] Optionally, the geological disaster emergency response knowledge graph includes a geological disaster sub-graph; the calculation formula for the mechanism similarity is:

[0033]

[0034] where E_similarity(T a , T b ) represents the mechanism similarity between the target geological disaster event T a and the historical geological disaster event T b , D a represents the set composed of the entities corresponding to the target geological disaster event T a in the geological disaster sub-graph, D b represents the set composed of the entities corresponding to the historical geological disaster event T b in the geological disaster sub-graph, |D a | represents the number of entities in the set D a , |D b | represents the number of entities in the set D b , k represents the number of entities matched by the set D a and the set D b , represents the entity corresponding to the attribute I of the target geological disaster event T a , represents the entity corresponding to the attribute I of the historical geological disaster event T b , represents and 's similarity value, w I represents the weight of the entity corresponding to the attribute I;

[0035] When the type of the entity I is a text type, 's calculation formula is:

[0036]

[0037] When the type of the entity I is a numerical type, 's calculation formula is:

[0038]

[0039] Optionally, screening the matching geological disaster events of the target geological disaster event from multiple reference geological disaster events based on graph similarity includes:

[0040] For each reference geological disaster event, based on the knowledge graph of geological disaster emergency response, calculate the comprehensive similarity between the target geological disaster event and the reference geological disaster event through the comprehensive similarity calculation formula;

[0041] Take the reference geological disaster event corresponding to the maximum comprehensive similarity as the matching geological disaster event of the target geological disaster event.

[0042] Optionally, the knowledge graph of geological disaster emergency response further includes a rescue environment sub-graph and an emergency response sub-graph; the comprehensive similarity calculation formula is:

[0043] similarity(T a ,T z )=Node_similarity(N a ,N z )+β×Edge_similarity(E a ,E z )

[0044] Among them, similarity(T a ,T z ) represents the comprehensive similarity between the target geological disaster event T a and the reference geological disaster event T z , N a represents the set of entities corresponding to the target geological disaster event T a in the geological disaster sub-graph and the rescue environment sub-graph, E a represents the set of relationships corresponding to the target geological disaster event T a in the geological disaster sub-graph and the rescue environment sub-graph, N z represents the set of entities corresponding to the reference geological disaster event T z in the geological disaster sub-graph and the rescue environment sub-graph, E z represents the set of relationships corresponding to the reference geological disaster event T z in the geological disaster sub-graph and the rescue environment sub-graph, Node_similarity(N a ,N z ) represents the node similarity between the target geological disaster event T a and the reference geological disaster event T z , β represents the weight coefficient, Edge_similarity(E a ,E z ) represents the edge similarity between the target geological disaster event T a and the reference geological disaster event T z .

[0045] Optionally, based on the geological disaster emergency response knowledge graph and the matching geological disaster events, an emergency response plan for the target geological disaster event is generated, including:

[0046] Extract entities that match the geological disaster events from the emergency response sub-graph;

[0047] Fill the extracted entities into a pre-set emergency response template to obtain an emergency response plan for the target geological disaster event.

[0048] Optionally, the geological disaster sub-graph is used to describe the basic attributes of geological disasters, the basic attributes of disaster events, and the knowledge of geological disaster mechanisms; the rescue environment sub-graph is used to describe the emergency response capabilities and rescue conditions; the emergency response sub-graph is used to describe the emergency response process.

[0049] The above solution of this application has the following beneficial effects:

[0050] In the embodiments of this application, by introducing a geological disaster mechanism constraint mechanism, it effectively avoids matching pseudo-similar geological disaster events under the interference of multiple attributes, ensures the accuracy of the matching of reference geological disaster events, and at the same time, through knowledge graph reasoning based on graph similarity, captures the explicit and potential relationships between attributes, and sets weights based on data, improving the matching accuracy of geological disaster events.

[0051] In addition, due to the high matching accuracy of geological disaster events, after the geological disaster events are matched, an emergency response plan with strong implementability can be automatically generated based on the matching results.

[0052] Other beneficial effects of this application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of this application, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the following drawings are only some embodiments of this application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0054] Figure 1 It is a flowchart of a method for generating a geological disaster emergency response plan based on a domain knowledge graph provided by an embodiment of this application;

[0055] Figure 2 It is a schematic diagram of the hierarchical structure of the geological disaster sub-graph in an example of this application;

[0056] Figure 3 It is a schematic diagram of the hierarchical structure of the rescue environment sub-graph in an example of this application;

[0057] Figure 4 Schematic diagram of the hierarchical structure of the emergency response sub-graph in an example of the present application;

[0058] Figure 5 Operation output diagram of the reference geological disaster event in an example of the present application;

[0059] Figure 6 Operation output diagram of the matching geological disaster event in an example of the present application. Detailed implementation manners

[0060] In the following description, for the purpose of illustration rather than limitation, specific details such as specific system structures and technologies are set forth in order to provide a thorough understanding of the embodiments of the present application. However, those skilled in the art should understand that the present application can be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary details.

[0061] It should be understood that when used in the specification of the present application and the appended claims, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0062] It should also be understood that the term "and / or" as used in the specification of the present application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0063] As used in the specification of the present application and the appended claims, the term "if" can be interpreted as "when", "once", "in response to determining", or "in response to detecting" according to the context. Similarly, the phrase "if determined" or "if [the described condition or event] is detected" can be interpreted as meaning "once determined", "in response to determining", "once [the described condition or event] is detected", or "in response to detecting [the described condition or event]" according to the context.

[0064] In addition, in the description of the specification of the present application and the appended claims, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0065] References to "one embodiment" or "some embodiments" in the description of the present application mean that a particular feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments of the present application. Thus, statements such as "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments", etc., which appear in different places in this specification, do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized. The terms "comprising", "including", "having" and their variants all mean "including but not limited to", unless otherwise specifically emphasized.

[0066] In view of the problems of low matching accuracy of current geological disaster events and poor feasibility of emergency response plans, the embodiments of the present application provide a method for generating a geological disaster emergency response plan based on a domain knowledge graph. By introducing a geological disaster mechanism constraint mechanism, this method effectively avoids matching pseudo-similar geological disaster events under the interference of multiple attributes, ensures the accuracy of matching reference geological disaster events, and at the same time, through knowledge graph reasoning based on graph similarity, captures explicit and potential relationships between attributes, and sets weights based on data, improving the matching accuracy of matching geological disaster events.

[0067] In addition, due to the high matching accuracy of matching geological disaster events, after determining the matching geological disaster events, an emergency response plan with strong feasibility can be automatically generated based on the matching geological disaster events.

[0068] The following uses specific embodiments to exemplarily illustrate the method for generating a geological disaster emergency response plan based on a domain knowledge graph of the present application.

[0069] As Figure 1 shown, the method for generating a geological disaster emergency response plan based on a domain knowledge graph provided by the embodiments of the present application includes the following steps:

[0070] Step 11: For each entity of the target geological disaster event in the pre-constructed geological disaster emergency response knowledge graph, determine the optimal path set of the entity.

[0071] The above-mentioned target geological disaster event is the geological disaster event for which an emergency response plan needs to be formulated currently; the above-mentioned optimal path set includes the optimal paths from each other entity in the geological disaster emergency response knowledge graph to the entity.

[0072] In some embodiments of the present application, the above-mentioned geological disaster emergency response knowledge graph includes a geological disaster sub-graph, a rescue environment sub-graph, and an emergency response sub-graph. Among them, the geological disaster sub-graph is used to describe the basic attributes of geological disasters, the basic attributes of disaster events, and the knowledge of geological disaster mechanisms; the rescue environment sub-graph is used to describe the emergency response capabilities and rescue conditions; the emergency response sub-graph is used to describe the emergency response process.

[0073] Specifically, the geological disaster emergency response knowledge graph can be constructed by integrating scientific and technological literature, historical monitoring data, reported danger data (i.e., historical geological disaster events and target geological disaster events), and real-time monitoring data. Specifically, the extraction and instantiation of multi-source knowledge of the knowledge graph can be completed based on various algorithms, formula calculations, and models.

[0074] In some embodiments of the present application, the above-mentioned geological disaster emergency response knowledge graph can be regarded as consisting of an ontology and its instantiated entities. Among them, the construction of the ontology such as disaster mechanisms, attributes, and characteristics is mainly based on scientific and technological literature, research reports, and monitoring data, covering the core entities of geological disasters and their relationships; the definition of entities and relationships related to emergency response and rescue environment mainly comes from emergency plans, action plans, online media, and relevant professional books, forming a complete emergency response knowledge system.

[0075] The data sources of different sub-graphs are diverse. According to the format characteristics of different data sources, corresponding technical means are adopted for processing:

[0076] Station monitoring data and professional spatial data: Through mapping nodes, the mapping extraction of the attributes, characteristics, and patterns of structured spatio-temporal data is completed, and combined with monitoring data (such as rainfall, groundwater level, InSAR deformation data), instance data of the geological disaster mechanism sub-graph is generated through formula calculation.

[0077] Scientific and technological literature, online texts, and investigation report data: Based on the Universal Information Extraction (UIE) model and the large model API, unstructured or semi-structured data is processed to extract instance data corresponding to the geological disaster mechanism graph.

[0078] Reported danger data: The reported data is converted into entities and relationships in the graph through mapping rules.

[0079] Emergency plans, action plans, and relevant professional books: Use OCR recognition, BERT ontology recognition, and OpenNRE relationship extraction to extract information, further extract entities and relationships related to emergency response, and construct instance data of the rescue environment and emergency response sub-graphs.

[0080] Finally, the MultiKE model is used to automatically align and fuse the entities and relationships in the triples. All data is ultimately stored in the Neo4j database, which contains 491 geological disaster events and their corresponding emergency response plans. The graph database contains 59,410 nodes and 39,609 edges, forming a knowledge graph in the field of geological disasters, namely, the geological disaster emergency response knowledge graph. The relationships of this graph are shown in Table 1.

[0081] Table 1

[0082]

[0083]

[0084] It can be understood that the entities in the geological disaster sub-graph include the basic attributes of geological disasters, the basic attributes of disaster events, and the knowledge of geological disaster mechanisms, such as disaster-causing factors (e.g., rainfall, earthquake). Exemplarily, taking the example of formulating an emergency response plan for a landslide disaster in a certain village, the hierarchical structure of the entities in the geological disaster sub-graph constructed by collecting relevant data can be as Figure 2 shown. In the figure, the geological disaster material characteristics, the basic attributes of disaster events, the geological disaster mechanism effects, and the basic attributes of geological disasters are the first-level entities (i.e., the first-tier entities). The subordinates of the first-level are the second-level entities (i.e., the second-tier entities), and so on. The figure also includes third-level entities, fourth-level entities, and fifth-level entities. It should be noted that due to space limitations, Figure 2 only part of the hierarchical entities of the geological disaster sub-graph are shown.

[0085] The geological disaster material characteristics include: landform characteristics, geological lithology characteristics, geological element characteristics, vegetation condition characteristics, hydrological characteristics, etc.

[0086] The basic attributes of disaster events: describe the basic information of disaster events.

[0087] The geological disaster mechanism effects include: natural effects (such as erosion, internal force, climate) and human effects (such as loading and unloading, irrigation, hydrodynamic, vibration).

[0088] The entities in the rescue environment sub-graph include emergency response capabilities, rescue conditions, such as material and equipment reserves, rescue site conditions, etc. Exemplarily, taking the example of formulating an emergency response plan for a landslide disaster in a certain village, the hierarchical structure of the entities in the rescue environment sub-graph constructed by collecting relevant data can be as Figure 3 shown. In the figure, human resource reserves, material and equipment reserves, training and drills, site conditions, road conditions, economic conditions, and social conditions are the first-level entities, and the subordinates of the first-level entities are the second-level entities. It should be noted that due to space limitations, Figure 3Only part of the hierarchical entities of the rescue environment sub-graph are shown.

[0089] Human resource reserves include: emergency management departments, fire departments, medical rescue departments, etc.

[0090] Material and equipment reserves include: personal protective equipment, search and rescue equipment, emergency transportation and professional operation transportation equipment, engineering machinery equipment, energy power equipment and materials, emergency lighting equipment and supplies, decontamination equipment and materials, logistics support equipment, non-powered hand tools, communication equipment, etc.

[0091] Training and drills include: training drill scenario facilities, training subject quality, number of drills, etc.

[0092] Site conditions include: layout conditions of large equipment, site width, hardness, flatness, etc.

[0093] Road conditions include: road traffic conditions.

[0094] Economic conditions include: urban GDP, insurance coverage, financial budget, etc.

[0095] Social conditions include: cultural factors, public awareness, volunteer network, etc.

[0096] The entities of the emergency response sub-graph include the actual process of emergency response, and the content is derived from emergency response plans and schemes. Of course, this emergency response sub-graph also includes the relationships between entities. Exemplarily, taking the formulation of an emergency response plan for a landslide disaster in a certain village as an example, the hierarchical structure of the entities in the emergency response sub-graph constructed by collecting relevant data can be as Figure 4 shown. In the figure, the emergency response system, rescue camps, material and equipment, and rescue teams are the first-level entities, and the subordinates of the first-level entities are the second-level entities. It should be noted that due to space limitations, Figure 4 only part of the hierarchical entities of the emergency response sub-graph are shown.

[0097] The emergency response system includes: Organization Department, Information Support Department, Medical Department, Rescue Department, etc.

[0098] The rescue camps include: the location of the rescue camp for resettlement, medical support materials, basic living material support, etc.

[0099] The material and equipment include: personal protective equipment, search and rescue equipment, emergency transportation and professional operation transportation equipment, engineering machinery equipment, energy power equipment and materials, emergency lighting equipment and supplies, decontamination equipment and materials, logistics support equipment, non-powered hand tools, communication equipment, etc.

[0100] The rescue teams include: the number of rescue team members, the number of rescue equipment, the types of rescue teams, etc.

[0101] It should be noted that in addition to the entities introduced above, the above geological disaster emergency response knowledge graph should also include the relationships between entities. Specifically, this knowledge graph dynamically connects the geological disaster mechanism, rescue environment constraints, and emergency response actions by defining the logical association relationships between entities. The construction of relationships is strictly based on geological action mechanisms (such as mechanical models and disaster chain evolution), emergency response process rules (such as pre-plan execution conditions), and environmental resource constraint conditions (such as site and equipment matching rules), and is deeply bound to the entity hierarchical structure (from level one to level five entities). Possible relationships include direct trigger, indirect trigger, interaction, collaboration, enhancement, destruction, correlation, etc.

[0102] For each entity of the target geological disaster event in the above geological disaster emergency response knowledge graph, the specific implementation method for determining the optimal path set of the entity is as follows:

[0103] For each other entity in the geological disaster emergency response knowledge graph except the entity, the optimal path from the other entity to the entity is determined through the path optimization calculation formula.

[0104] The path optimization calculation formula is:

[0105] PathBA = arg P∈Path_Set(B,A) max(F(P))

[0106] F(P) = IIS(P) / L(P)

[0107] IIS(P) = S(r B1 ) × S(r 12 ) × … S(r nA )

[0108] L(P) = |V| - 1

[0109] Among them, PathBA represents the optimal path from other entity B to entity A, Path_Set(B,A) represents the set of all paths from other entity B to entity A, IIS(P) represents the path semantic intensity of path P in Path_Set(B,A), the path semantic intensity can be preset, specifically it can be set based on literature statistics and expert scoring method, L(P) represents the path length of path P, S(r B1 ) represents the semantic intensity of relationship r B1 in path P, S(r 12 ) represents the semantic intensity of relationship r 12 in path P, S(r nA ) represents the semantic intensity of relationship r nA in path P, the semantic intensity can be preset, specifically it can be set based on literature statistics and expert scoring method, r B1 represents the relationship between other entity B and entity e 1The relationship, r 12 represents entity e 1 and entity e 2 's relationship, r nA represents represents entity e n 's relationship with entity A, entity e 1 , e 2 and e n are entities on path P, and |V| represents the number of entities included in path P.

[0110]

[0111] P = {(B, e 1 , r B1 ), (e 1 , e 2 , r 12 ), … (e n , A, r nA )}, where E is the set of all entities, e 1 , e 2 represent entities, R is the set of relationships involving influence, for example R = {synergy, strengthening}, Rel(e 1 , e 2 , r 12 ) represents that e 1 is connected to e 12 through relationship r 2 , r B1 represents the relationship between another entity B and entity e 1 , and Path(B, A, P) represents the path P from another entity B to entity A, and P is a set of a series of relationships and entities.

[0112] Taking the formulation of an emergency response plan for a landslide disaster in a certain village as an example (i.e., the target geological disaster event is the landslide disaster in a certain village), the semantic intensities of some relationships are shown in Table 2:

[0113] Table 2

[0114]

[0115] To avoid causal fallacies and ensure that only one single path is retained between each pair of entities, it is necessary to query all entities and the set of optimal paths. For entity A of the target geological disaster event and other entity B in the geological disaster emergency response knowledge graph, if there are both the optimal path PathAB from entity A to other entity B and the optimal path PathBA from other entity B to entity A, then the optimal path is retained through the following formula:

[0116]

[0117] Among them, Level(A) represents the level of entity A in the knowledge graph of geological disaster emergency response, Level(B) represents the level of other entity B in the knowledge graph of geological disaster emergency response, TC(PathAB) represents the path smoothness of PathAB, TC(PathBA) represents the path smoothness of PathBA, IIS(PathAB) represents the path semantic intensity of PathAB, exp(-L(PathAB)) represents the path length attenuation factor of PathAB, IIS(PathBA) represents the path semantic intensity of PathBA, and exp(-L(PathBA)) represents the path length attenuation factor of PathBA. The path semantic intensity and the path length attenuation factor can be preset, specifically based on literature statistics and expert scoring methods.

[0118] It can be understood that after selecting between the optimal path PathAB and the optimal path PathBA, it is necessary to update the optimal path sets of entity A and other entity B.

[0119] It should be noted that the above entity A and other entity B can be any two entities in the knowledge graph of geological disaster emergency response.

[0120] Step 12: Determine the weight of each entity of the target geological disaster event in the knowledge graph of geological disaster emergency response according to the optimal path set of each entity.

[0121] In some embodiments of the present application, for each entity of the target geological disaster event in the knowledge graph of geological disaster emergency response, the weight w of the entity can be calculated through the following formula A :

[0122]

[0123] Among them, w 0 represents the initial weight, the initial weight can be set to 1, Path_Set(A) represents the optimal path set of entity A, e i represents the i-th entity on the optimal path Q, e i+1 represents the (i + 1)-th entity on the optimal path Q, r i(i+1) represents the relationship between e i and e i+1 , S(r i(i+1) ) represents the semantic intensity of r i(i+1) , represents the distance attenuation factor of the i-th entity e i on the optimal path Q, d(e i , A) represents the number of hops from the i-th entity e i on the optimal path Q to entity A, and this number of hops refers to the entity e iThe number of entities between entity A.

[0124] The above semantic intensity and distance attenuation factor can be preset, and can be specifically set based on literature statistics and expert scoring method.

[0125] Taking the formulation of an emergency response plan for a landslide disaster in a certain village as an example (i.e., the target geological disaster event is the landslide disaster in a certain village), part of the weight adjustment results are shown in Table 3:

[0126] Table 3

[0127]

[0128]

[0129] Step 13, with the similarity of geological disaster mechanism action as a constraint, based on the determined weights, select multiple reference geological disaster events of the target geological disaster event from the historical geological disaster events used when constructing the geological disaster emergency response knowledge graph.

[0130] The entities of the above geological disaster emergency response knowledge graph correspond to the attributes of historical geological disaster events and target geological disaster events (that is, the entities of the geological disaster emergency response knowledge graph include the attributes of historical geological disaster events and the attributes of target geological disaster events, and the attributes of historical geological disaster events and the attributes of target geological disaster events are also the entities of the geological disaster emergency response knowledge graph). The attributes of historical geological disaster events and the attributes of target geological disaster events can both be understood as including some or all of the entities such as geological disaster material characteristics, basic attributes of disaster events, geological disaster mechanism action, basic attributes of geological disasters, human resource reserves, material and equipment reserves, training and drills, site conditions, road conditions, economic conditions, social conditions, emergency systems, rescue camps, material and equipment, rescue teams, etc.

[0131] In some embodiments of the present application, for each historical geological disaster event used when constructing the geological disaster emergency response knowledge graph, the mechanism similarity between the target geological disaster event and the historical geological disaster event can be calculated through the mechanism similarity calculation formula; then the historical geological disaster event corresponding to the mechanism similarity with a value greater than the preset similarity threshold is used as the reference geological disaster event of the target geological disaster event.

[0132] The above target geological disaster event is the geological disaster event for which an emergency response plan needs to be formulated currently, and the above preset similarity threshold can be set according to the actual situation.

[0133] It should be noted that when screening the reference geological disaster events, it is mainly achieved by using the geological disaster sub-graph in the geological disaster emergency response knowledge graph. The above mechanism similarity calculation formula is:

[0134]

[0135] Among them, E_similarity(T a , T b ) represents the mechanism similarity between the target geological disaster event T a and the historical geological disaster event T b . D a represents the set composed of the entities corresponding to the target geological disaster event T a in the geological disaster sub-graph. D b represents the set composed of the entities corresponding to the historical geological disaster event T b in the geological disaster sub-graph. |D a | represents the number of entities in the set D a . |D b | represents the number of entities in the set D b . k represents the number of entities matched by the set D a and the set D b (that is, the number of entities included in both the set D a and the set D b ). represents the entity corresponding to the attribute I of the target geological disaster event T a . represents the entity corresponding to the attribute I of the historical geological disaster event T b . represents the similarity value between . w I represents the weight of the entity corresponding to the attribute I.

[0136] When the type of the entity I is a text type, the calculation formula of

[0137]

[0138] When the type of the entity I is a numerical type, the calculation formula of

[0139]

[0140] If then is 1.

[0141] Taking the formulation of an emergency response plan for the landslide disaster in a certain village as an example (that is, the target geological disaster event is the landslide disaster in a certain village), the information of the reference geological disaster event output is as Figure 5 shown. In the figure, the first column is the number of the reference geological disaster event.

[0142] Step 14: Screen out the matching geological disaster events of the target geological disaster event from multiple reference geological disaster events based on graph similarity.

[0143] In some embodiments of the present application, for each reference geological disaster event, based on the geological disaster emergency response knowledge graph, the comprehensive similarity between the target geological disaster event and the reference geological disaster event can be calculated through a comprehensive similarity calculation formula; then the reference geological disaster event corresponding to the largest comprehensive similarity value is used as the matching geological disaster event of the target geological disaster event.

[0144] It should be noted that when screening the matching geological disaster events, the rescue environment sub-graph and the emergency response sub-graph in the geological disaster emergency response knowledge graph are mainly used to implement. The above comprehensive similarity calculation formula is:

[0145] similarity(T a ,T z )=Node_similarity(N a ,N z )+β×Edge_similarity(E a ,E z )

[0146] Among them, similarity(T a ,T z ) represents the comprehensive similarity between the target geological disaster event T a and the reference geological disaster event T z , N a represents the set composed of the entities corresponding to the target geological disaster event T a in the geological disaster sub-graph and the rescue environment sub-graph, E a represents the set composed of the relationships corresponding to the target geological disaster event T a in the geological disaster sub-graph and the rescue environment sub-graph, N z represents the set composed of the entities corresponding to the reference geological disaster event T z in the geological disaster sub-graph and the rescue environment sub-graph, E z represents the set composed of the relationships corresponding to the reference geological disaster event T z in the geological disaster sub-graph and the rescue environment sub-graph, Node_similarity(N a ,N z ) represents the similarity between the target geological disaster event T a and the reference geological disaster event T zThe node similarity between them, β represents the weight coefficient, which can be set according to experience. Edge_similarity(E a ,E z ) represents the edge similarity between the target geological disaster event T a and the reference geological disaster event T z .

[0147] Among them, the node similarity can be calculated using the mechanism similarity calculation formula: Among them, D a is N a , D b is N z , the historical geological disaster event T b is the reference geological disaster event T z . And it should be noted that the node refers to an entity, so there are both numerical types and text types when calculating the node similarity.

[0148] The edge similarity can be calculated using the following formula:

[0149]

[0150] Among them, r a and r z respectively represent the edges of the target geological disaster event T a and the reference geological disaster event T z corresponding to the same type r. k represents the number of edges matched by the set E a and the set E z (that is, the number of edges included in both the set E a and the set E z ), r a represents the edge of the corresponding relationship r of the target geological disaster event T a (for example, r is the relationship between the entity rainfall and the entity landslide. In the target geological disaster event T a , r a is a direct trigger), r z represents the edge of the corresponding relationship r of the reference geological disaster event T z (for example, r is the relationship between the entity rainfall and the entity landslide. In the reference geological disaster event T z , r z is an indirect trigger), sim(r a ,r z ) represents the similarity value of r a and r b .

[0151] Taking the formulation of an emergency response plan for a mountain landslide disaster in a certain village as an example (that is, the target geological disaster event is a mountain landslide disaster in a certain village), the output matching geological disaster event information is as Figure 6As shown in the figure. In the figure, the first column is the number of historical geological disaster events (where the historical geological disaster event numbered 0 is the matching geological disaster event).

[0152] Step 15, generate an emergency response plan for the target geological disaster event based on the geological disaster emergency response knowledge graph and the matching geological disaster event.

[0153] In some embodiments of the present application, entities of the matching geological disaster event can be extracted from the emergency response sub-graph; then the extracted entities are filled into a pre-set emergency response template to obtain an emergency response plan for the target geological disaster event.

[0154] The above-mentioned extracted entities can be the specific contents of emergency systems, rescue camps, material equipment, rescue teams, etc. The emergency response template contains the basic structure and content of the emergency response plan, but there are some specific parts in this template that are dynamic, that is, they will change according to specific disaster events. These parts are the so-called "dynamic data points". In order to be able to automatically fill these dynamic data points, specific "placeholder" are used to mark them in the template. The placeholder is like a blank in the template, waiting to be filled with specific information. Specifically, after the above entities are extracted, these entities can be filled into the placeholders in the template, and this process can be achieved through an automated "text template engine". The text template engine is a software tool that can identify the placeholders in the template and fill in the corresponding data. Finally, the text template engine will generate a complete and formatted emergency response plan document. This document already contains all the necessary information and can be directly used to guide emergency response actions.

[0155] In practical applications, the emergency response sub-graph of the target geological disaster event can also be improved and updated according to the geological disaster event. Specifically as follows: First, determine whether each ontology in the emergency response sub-graph has an entity associated with the entity of the target geological disaster event. If an ontology does not have an entity associated with the target geological disaster event, record this ontology and add its name to the set of ontologies to be instantiated. Traverse the matching geological disaster events. If there are entities that meet the set of ontologies to be instantiated, copy the entities and associate them with the target geological disaster event.

[0156] The above is the preferred implementation manner of the present application. It should be noted that for those of ordinary skill in the art in this technical field, without departing from the principle described in the present application, several improvements and refinements can still be made, and these improvements and refinements should also be regarded as the protection scope of the present application.

Claims

1. A method for generating geological disaster emergency response plans based on domain knowledge graph, characterized in that: include: For each entity of the target geological disaster event in the pre-constructed geological disaster emergency response knowledge graph, determine the optimal path set of the entity; the optimal path set includes the optimal path from each other entity in the geological disaster emergency response knowledge graph to the entity; Determine the weight of each entity of the target geological disaster event in the geological disaster emergency response knowledge graph according to the optimal path set of each entity; Taking the similarity of the mechanism of geological disasters as a constraint, multiple reference geological disaster events of the target geological disaster event are screened out from the historical geological disaster events used in constructing the geological disaster emergency response knowledge graph based on the determined weights; the entities of the geological disaster emergency response knowledge graph correspond to the attributes of the historical geological disaster events and the target geological disaster event; Screening out matching geological disaster events of the target geological disaster event from the multiple reference geological disaster events based on graph similarity; Based on the geological disaster emergency response knowledge graph and the matching geological disaster events, an emergency response plan for the target geological disaster event is generated.

2. The method for generating a geological disaster emergency response plan according to claim 1, characterized in that: The determining of the optimal path set of the entity comprises: For each entity other than the entity in the geological disaster emergency response knowledge graph, respectively, determine the optimal path from the other entity to the entity through a path optimization calculation formula; The path optimization calculation formula is: PathBA=arg P∈Path_Set(B,A) max(F(P)) F(P)=IIS(P) / L(P) IIS(P)=S(r B1 )×S(r 12 )×…S(r nA ) L(P)=|V|-1 Where PathBA represents the optimal path from other entity B to entity A, Path_Set(B,A) represents the set of all paths from other entity B to entity A, IIS(P) represents the path semantic strength of path P in Path_Set(B,A), L(P) represents the path length of path P, S(r B1 ) represents the relationship r in path P B1 The semantic strength of S(r 12 ) represents the relationship r in path P 12 The semantic strength of S(r nA ) represents the relationship r in path P nA The semantic strength of B1 Indicates the relationship between other entities B and entity e1, r 12 Represents the relationship between entity e1 and entity e2, r nA Represents entity e n The relationship between entity A and entity e1, e2 and e n is the entity on path P, and |V| represents the number of entities contained in path P.

3. The method for generating a geological disaster emergency response plan according to claim 2, characterized in that: The method for generating a geological disaster emergency response plan also includes: For the entity A and other entities B of the target geological disaster event in the geological disaster emergency response knowledge graph, if there is an optimal path PathAB from the entity A to other entities B and an optimal path PathBA from other entities B to the entity A, the optimal path is retained by the following formula: Among them, Level(A) represents the level of entity A in the geological disaster emergency response knowledge graph, Level(B) represents the level of other entities B in the geological disaster emergency response knowledge graph, TC(PathAB) represents the path smoothness of PathAB, TC(PathBA) represents the path smoothness of PathBA, IIS(PathAB) represents the path semantic strength of PathAB, exp(-L(PathAB)) represents the path length attenuation factor of PathAB, IIS(PathBA) represents the path semantic strength of PathBA, and exp(-L(PathBA)) represents the path length attenuation factor of PathBA.

4. The method for generating a geological disaster emergency response plan according to claim 3, characterized in that: Determining the weight of each entity of the target geological disaster event in the geological disaster emergency response knowledge graph according to the optimal path set of each entity includes: For each entity of the target geological disaster event in the geological disaster emergency response knowledge graph, the weight w of the entity is calculated by the following formula: A : Wherein, w0 represents the initial weight, Path_Set(A) represents the optimal path set of the entity A, e i represents the i-th entity on the optimal path Q, e i+1 represents the i+1th entity on the optimal path Q, r i(i+1) Indicates e i With e i+1 The relationship between S(r i(i+1) ) represents r i(i+1) The semantic strength of Represents the i-th entity e on the optimal path Q i The distance attenuation factor, d(e i ,A) represents the i-th entity e on the optimal path Q i The number of hops to the entity A.

5. The method for generating a geological disaster emergency response plan according to claim 4, characterized in that: The method uses the similarity of the mechanism of geological disasters as a constraint and based on the determined weights, selects multiple reference geological disaster events of the target geological disaster event from the historical geological disaster events used in constructing the geological disaster emergency response knowledge graph, including: For each historical geological disaster event used in constructing the geological disaster emergency response knowledge graph, the mechanism similarity between the target geological disaster event and the historical geological disaster event is calculated by using a mechanism similarity calculation formula; The historical geological disaster events corresponding to the mechanism similarity whose values ​​are greater than the preset similarity threshold are used as reference geological disaster events for the target geological disaster events.

6. The method for generating a geological disaster emergency response plan according to claim 5, characterized in that: The geological disaster emergency response knowledge graph includes a geological disaster sub-graph; the mechanism similarity calculation formula is: Among them, E_similarity(T a ,T b ) represents the target geological disaster event T a The historical geological disaster events b The similarity of the mechanism between a Represents the target geological disaster event T described in the geological disaster sub-map a The corresponding entity set, D b Represents the historical geological disaster events T described in the geological disaster sub-atlas b The corresponding entity set, |D a | represents the set D a Number of entities in |D b | represents the set D b The number of entities in the set D a and set D b The number of matching entities, Represents the target geological disaster event T a The entity corresponding to the attribute I, Represents the historical geological disaster event T b The entity corresponding to the attribute I, express and The similarity value, w I Represents the weight of the entity corresponding to attribute I; When the type of entity I is text type, The calculation formula is: When the type of entity I is a numeric type, The calculation formula is:

7. The method for generating a geological disaster emergency response plan according to claim 6, characterized in that: The step of selecting a matching geological disaster event of the target geological disaster event from the plurality of reference geological disaster events based on graph similarity includes: For each reference geological disaster event, based on the geological disaster emergency response knowledge graph, the comprehensive similarity between the target geological disaster event and the reference geological disaster event is calculated by a comprehensive similarity calculation formula; The reference geological disaster event corresponding to the comprehensive similarity with the largest value is used as the matching geological disaster event of the target geological disaster event.

8. The method for generating a geological disaster emergency response plan according to claim 7, characterized in that: The geological disaster emergency response knowledge graph also includes a rescue environment sub-graph and an emergency response sub-graph; the comprehensive similarity calculation formula is: similarity(T a ,T z )=Node_similarity(N a ,N z )+β×Edge_similarity(E a ,E z ) Among them, similarity(T a ,T z ) represents the target geological disaster event T a Compared with the reference geological disaster event T z The comprehensive similarity between a Represents the target geological disaster event T described in the geological disaster sub-map and the rescue environment sub-map a The corresponding entity set, E a Represents the target geological disaster event T in the geological disaster sub-map and the rescue environment sub-map a The set of corresponding relations, N z Represents the reference geological disaster event T described in the geological disaster sub-map and the rescue environment sub-map z The corresponding entity set, E z Represents the reference geological disaster event T in the geological disaster sub-map and the rescue environment sub-map z The corresponding relationship consists of a set of Node_similarity(N a ,N z ) represents the target geological disaster event T a Compared with the reference geological disaster event T z The node similarity between them, β represents the weight coefficient, Edge_similarity(E a ,E z ) represents the target geological disaster event T a Compared with the reference geological disaster event T z The edge similarity between .

9. The method for generating a geological disaster emergency response plan according to claim 8, characterized in that: The generating of the emergency response plan for the target geological disaster event based on the geological disaster emergency response knowledge graph and the matching geological disaster event includes: Extracting the entity matching the geological disaster event from the emergency response sub-graph; The extracted entities are filled into a pre-set emergency response template to obtain an emergency response plan for the target geological disaster event.

10. The method for generating a geological disaster emergency response plan according to claim 8, characterized in that: The geological disaster sub-map is used to describe the basic attributes of geological disasters, the basic attributes of disaster events and the knowledge of geological disaster mechanisms; the rescue environment sub-map is used to describe emergency response capabilities and rescue conditions; and the emergency response sub-map is used to describe the emergency response process.

Images