Network construction and assessment method and device for urban flood cascade compound disaster events

Abstract

The present invention discloses a method and device for constructing and evaluating an urban flood cascade composite disaster event network, comprising the steps of obtaining a plurality of initial urban flood disaster text data, pre-processing each initial urban flood disaster text data, and determining a plurality of target urban flood disaster text data; extracting causal relationships from each target urban flood disaster text data, and generating a plurality of explicit event text data and a plurality of explicit and implicit event text data; parsing each explicit event text data and each explicit and implicit event text data and performing event fusion to determine a plurality of target event text data; constructing a target urban flood cascade composite disaster event network based on each target event text data; and evaluating the target urban flood cascade composite disaster event network to generate urban flood disaster assessment data. The method solves the technical problem that existing urban flood network construction and risk assessment technologies are difficult to comprehensively and effectively assess the risks brought about by urban flood disasters.

Description

Technical Field

[0001] This invention relates to the field of event graph construction technology, and in particular to a method and apparatus for constructing and evaluating a network of cascading complex disaster events related to urban flooding. Background Technology

[0002] Under the influence of climate change and human activities, urban flooding has become increasingly frequent in recent years. It not only directly causes large-scale waterlogging but also leads to natural geological disasters such as landslides and mudslides, as well as accidents such as building collapses, road and infrastructure damage, and dam breaches, resulting in severe economic losses and casualties. The development of urban flooding disasters exhibits a significant chain-like transmission characteristic, often accompanied by cascading, superimposed, and amplified effects.

[0003] Because urban flooding involves both natural and human systems, its impacts are complex, intrinsically correlated, and nonlinear. Research shows that it is feasible to apply complex network theory to represent and analyze the complexity and nonlinearity of flood disaster systems. Complex networks are topological structures composed of nodes and edges connecting them, typically possessing one or more properties such as small-world, clustering, and scale-free characteristics, making them suitable for characterizing complex systems in the real world.

[0004] Existing urban flood network construction and risk assessment technologies mostly study single disaster events, neglecting the interaction between flood-related disaster events. This is not conducive to revealing the development patterns of cascading and complex urban flood disaster events, and makes it difficult to conduct a comprehensive and effective assessment of the risks posed by urban flood disasters. Summary of the Invention

[0005] This invention provides a method and apparatus for constructing and assessing a network of cascading and complex urban flood disasters, which addresses the technical problem that existing urban flood network construction and risk assessment technologies are unable to comprehensively and effectively assess the risks posed by urban flood disasters.

[0006] The first aspect of this invention provides a method for constructing and evaluating a network of cascading and complex urban flood disaster events, comprising:

[0007] Multiple initial urban flood disaster text data are obtained, and each initial urban flood disaster text data is preprocessed to determine multiple target urban flood disaster text data;

[0008] The causal relationship was extracted from the flood disaster text data of each target city using regular expression method and pre-set causal extraction model, generating multiple explicit event text data and multiple explicit and implicit event text data;

[0009] The explicit event text data and the explicit and implicit event text data are analyzed and fused to determine multiple target event text data.

[0010] A network of cascading complex flood disaster events in the target city was constructed using graph database visualization software based on the text data of each target event.

[0011] A pre-defined risk assessment function is used to evaluate the network of cascading and complex flood disaster events in the target city, generating urban flood disaster assessment data.

[0012] Optionally, the step of preprocessing the initial urban flood disaster text data to determine multiple target urban flood disaster text data includes:

[0013] The initial urban flood disaster text data are deduplicated to generate deduplicated urban flood disaster text data;

[0014] Denoising operations are performed on the deduplicated urban flood disaster text data to generate denoised urban flood disaster text data;

[0015] Based on a pre-set place name extraction model, latitude and longitude filtering is performed on the denoised urban flood disaster text data to determine multiple target city flood disaster text data.

[0016] Optionally, the explicit event text data includes explicit causal relationships, explicit cause events, and explicit result events; the explicit and implicit event text data includes explicit and implicit causal relationships, explicit and implicit cause events, and explicit and implicit result events; the step of extracting causal relationships from the flood disaster text data of each target city using regular expressions and a pre-set causal extraction model to generate multiple explicit event text data and multiple explicit and implicit event text data includes:

[0017] The regular expression method is used to extract explicit causal relationships from the text data of flood disasters in the target cities, and outputs multiple explicit causal relationships, multiple explicit cause events, and multiple explicit result events;

[0018] A pre-defined causal extraction model is used to extract explicit and implicit causal relationships from the textual data of flood disasters in the target cities, generating multiple explicit and implicit causal relationships, multiple explicit and implicit cause events, and multiple explicit and implicit result events.

[0019] Optionally, the target event text data includes target cause event text data and target result event text data; the step of parsing each of the explicit event text data and each of the explicit and implicit event text data and performing event fusion to determine multiple target event text data includes:

[0020] The main body of each explicit cause event and each explicit result event is extracted to generate multiple first cause event representation information and multiple first result event representation information;

[0021] The main body of each explicit and implicit cause event and each explicit and implicit result event is extracted to generate multiple second cause event representation information and multiple second result event representation information;

[0022] Event fusion is performed on each of the first cause event representation information, each of the second cause event representation information, each of the first result event representation information, and each of the second result event representation information to generate target cause event text data and target result event text data.

[0023] Optionally, the graph database visualization software includes a graph database and complex network analysis software; the step of using the graph database visualization software to construct a target city flood cascading complex disaster event network based on the text data of each target event includes:

[0024] Based on a graph database, the text data of each target cause event and the text data of each target result event are used as disaster event nodes, and the explicit and implicit causal relationships and the explicit causal relationships are used as directed edges to construct an initial urban flood cascaded complex disaster event network.

[0025] The initial urban flood cascade complex disaster event network was visualized using complex network analysis software to generate the target city flood cascade complex disaster event network.

[0026] Optionally, the step of evaluating the target city's cascading complex disaster event network using a pre-set risk assessment function to generate urban flood disaster assessment data includes:

[0027] Extract disaster event node data and directed edge data from the target city's flood cascade complex disaster event network;

[0028] Substitute the disaster event node data and the directed edge data into a preset risk assessment function to calculate urban flood disaster assessment data.

[0029] The second aspect of this invention provides a device for constructing and evaluating a network of cascading and complex urban flood disaster events, comprising:

[0030] The acquisition module is used to acquire multiple initial urban flood disaster text data, and preprocess each of the initial urban flood disaster text data to determine multiple target urban flood disaster text data;

[0031] A module is used to extract causal relationships from the flood disaster text data of each target city using regular expressions and a pre-built causal extraction model, generating multiple explicit event text data and multiple explicit and implicit event text data.

[0032] The parsing module is used to parse the explicit event text data and the explicit and implicit event text data and perform event fusion to determine multiple target event text data;

[0033] The construction module is used to construct a network of cascading complex flood disaster events in the target city based on the text data of each target event using graph database visualization software;

[0034] The generation module is used to evaluate the target city's flood cascade complex disaster event network using a pre-set risk assessment function, and generate urban flood disaster assessment data.

[0035] A third aspect of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor causes the processor to perform the steps of the method for constructing and evaluating urban flood cascading complex disaster event networks as described in any of the preceding claims.

[0036] The fourth aspect of the present invention provides a computer-readable storage medium having a computer program / instructions stored thereon, wherein the computer program / instructions, when executed by a processor, implement the steps of the method for constructing and evaluating a network of cascading complex disaster events in urban flooding as described in any of the preceding claims.

[0037] The fifth aspect of the present invention provides a computer program product, comprising a computer program / instructions, which, when executed by a processor, implement the steps of the method for constructing and evaluating a network of cascading complex disaster events in urban flooding as described in any of the preceding claims.

[0038] As can be seen from the above technical solutions, the present invention has the following advantages:

[0039] The above-mentioned technical solution of the present invention provides a method for constructing and evaluating a network of cascading complex disaster events in urban flooding. First, multiple initial urban flooding disaster text data are acquired and preprocessed to determine multiple target urban flooding disaster text data. Then, a regular expression method and a pre-set causal extraction model are used to extract causal relationships from the target urban flooding disaster text data, generating multiple explicit event text data and multiple explicit / implicit event text data. The explicit and implicit event text data are analyzed and fused to determine multiple target event text data. A graph database visualization software is used to construct a network of cascading complex disaster events in the target cities based on the target event text data. Finally, a pre-set risk assessment function is used to evaluate the network of cascading complex disaster events in the target cities. The process involves estimating and generating urban flood disaster assessment data. Based on the above scheme, multiple explicit and implicit event text data are obtained using regular expression methods and a pre-set causal extraction model. These explicit and implicit event text data are then analyzed and fused. Using graph database visualization software, the obtained target event text data are used to construct a target city flood cascade complex disaster event network. This process considers the interaction between flood-induced disaster events, which helps reveal the development patterns of urban flood cascade complex disaster events. Furthermore, by using a pre-set risk assessment function to evaluate the target city flood cascade complex disaster event network and generate urban flood disaster assessment data, a comprehensive and effective assessment of the disaster event network can be achieved, further reducing the risks posed by urban flood disasters. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 A flowchart illustrating the steps of a method for constructing and evaluating a network of cascading complex disaster events in urban flooding, as provided in an embodiment of the present invention;

[0042] Figure 2 A schematic diagram illustrating the input and output examples of the BERT-BiLSTM-CRF model provided in this embodiment of the invention;

[0043] Figure 3 A flowchart illustrating the steps of another method for constructing and evaluating a network of cascading urban flood disaster events provided in this embodiment of the invention;

[0044] Figure 4A flowchart illustrating the method for constructing and evaluating a network of cascading and complex urban flood disaster events provided in an embodiment of the present invention;

[0045] Figure 5 This is a structural block diagram of a device for constructing and evaluating a network of cascading complex disaster events in urban flooding, provided as an embodiment of the present invention. Detailed Implementation

[0046] This invention provides a method and apparatus for constructing and assessing a network of cascading and complex urban flood disasters, which addresses the technical problem that existing urban flood network construction and risk assessment technologies are insufficient to comprehensively and effectively assess the risks posed by urban flood disasters.

[0047] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0048] This invention refers to urban flooding and the resulting natural disasters and accidents as urban flooding cascaded complex disaster events. Using social media (Sina Weibo) data as the data source, this invention proposes a method for constructing and evaluating the network of urban flooding cascaded complex disaster events from a network perspective.

[0049] Please see Figure 1 , Figure 1 A flowchart illustrating the steps of a method for constructing and evaluating a network of cascading complex disaster events in urban flooding, as provided in an embodiment of the present invention.

[0050] This invention provides a method for constructing and evaluating a network of cascading and complex urban flood disaster events, comprising:

[0051] Step 101: Obtain multiple initial urban flood disaster text data, preprocess each initial urban flood disaster text data, and determine multiple target urban flood disaster text data.

[0052] It should be noted that the initial urban flood disaster text data was obtained using Python web crawling technology based on XPath. The specific acquisition process can be divided into five steps: 1) User logs in to Weibo and obtains the webpage cookie; 2) The Request library is used to obtain the webpage source code; 3) Retrieve relevant Weibo posts using flood disaster keywords; 4) Analyze the HTML source code using the lxml library; 5) Obtain Weibo text, time, and other data. The flood disaster keywords are shown in Table 1.

[0053] Table 1 Keywords related to flood disasters

[0054]

[0055] Furthermore, since the Weibo data (initial urban flood disaster text data) obtained based on keywords contains duplicates and data unrelated to flood disasters, it is necessary to preprocess each initial urban flood disaster text data. Specifically, the initial urban flood disaster text data is deduplicated to generate deduplicated urban flood disaster text data, and the deduplicated urban flood disaster text data is then denoised to generate denoised urban flood disaster text data.

[0056] Furthermore, to obtain data within the study area more accurately, it is necessary to filter the data based on place names in the Weibo text. Specifically, based on a pre-set place name extraction model, the latitude and longitude of the denoised urban flood disaster text data are filtered to determine multiple target urban flood disaster text data. Specifically, the place names of the denoised urban flood disaster text data are extracted using the pre-set place name extraction model, and the place name data corresponding to each denoised urban flood disaster text data is output. Then, the place name data is converted into corresponding latitude and longitude through the Gaode Map Open Platform. The latitude and longitude of the selected latitude and longitude are filtered to be located in the pre-set study area, and the denoised urban flood disaster text data corresponding to the selected latitude and longitude are used as the target urban flood disaster text data. The pre-set study area can be Guangzhou, Shanghai, etc., and those in this technical field can set it as needed, without specific limitations here.

[0057] Further, please refer to Figure 2 The pre-built place name extraction model is BERT-BiLSTM-CRF (Bidirectional Encoder Representation from Transformers-Bidirectional Long ShortTerm Memory-Conditional Random Fields), used to extract place names from Weibo text data. This model is widely used in the field of named entity recognition and has advantages such as high accuracy and strong generalization ability. This invention uses the People's Daily dataset for model training and testing, with BIO annotation format. The training, validation, and test sets contain 20865, 2318, and 4636 sentences, respectively. Examples of the input and output of the trained model are shown below. Figure 2 As shown. By using the API (Application Programming Interface) provided by the Gaode Map Open Platform to convert the place names output by the model into latitude and longitude, data within the pre-defined study area can be filtered based on latitude and longitude.

[0058] In this embodiment, multiple initial urban flood disaster text data are acquired, and each initial urban flood disaster text data is preprocessed to determine multiple target urban flood disaster text data.

[0059] Step 102: Use regular expressions and a pre-set causal extraction model to extract causal relationships from the flood disaster text data of each target city, generating multiple explicit event text data and multiple explicit and implicit event text data.

[0060] Regular expression matching is a pattern matching method using regular expressions. This method consists of a series of character sequences and special characters, used to describe a template of text patterns to be matched.

[0061] The pre-built causal extraction model is the ERNIE model (Enhanced Representation through kNowledgeIntEgration, a pre-trained language model). This model is a pre-trained model for continuous learning semantic understanding based on knowledge enhancement. It combines pre-training with rich knowledge from multiple sources and achieves continuous evolution of model performance through continuous learning technology.

[0062] Explicit event text data includes explicit causal relationships, explicit cause events, and explicit result events.

[0063] Explicit and implicit event text data includes explicit and implicit causal relationships, explicit and implicit cause events, and explicit and implicit result events.

[0064] It should be noted that causal relationships in text typically consist of three parts: causal events, causal relationship cue words, and result events. Causal relationship cue words are words and phrases that indicate the causal relationship between events. Based on whether both causal and result events appear simultaneously in the text, causal relationships are classified as implicit or explicit. Considering the characteristics of causal relationships, this invention employs a pattern matching method based on regular expressions to extract explicit causal relationships, and a deep learning method based on the ERNIE model developed by Baidu to extract both explicit and implicit causal relationships.

[0065] Specifically, the regular expression method is used to extract explicit causal relationships from the flood disaster text data of each target city, and output multiple explicit causal relationships, multiple explicit cause events, and multiple explicit result events; a pre-set causal extraction model is used to extract explicit and implicit causal relationships from the flood disaster text data of each target city, and generate multiple explicit and implicit causal relationships, multiple explicit and implicit cause events, and multiple explicit and implicit result events.

[0066] In this embodiment, regular expressions and a pre-set causal extraction model are used to extract causal relationships from the flood disaster text data of each target city, generating multiple explicit event text data and multiple explicit and implicit event text data.

[0067] Step 103: Analyze the text data of each explicit event and the text data of each explicit and implicit event, and perform event fusion to determine multiple target event text data.

[0068] Target event text data includes target cause event text data and target result event text data;

[0069] It should be noted that the causal events and result events (explicit causal events and explicit result events, explicit and implicit causal events and explicit and implicit result events) obtained through regular expression pattern matching and pre-built causal extraction models still contain redundant information. It is necessary to extract the core components representing the event information. Specifically, the core components are extracted for each explicit causal event and each explicit result event to generate the first causal event representation information and the first result event representation information corresponding to each explicit causal event. Similarly, the core components are extracted for each explicit and implicit causal event and each explicit and implicit result event to generate the second causal event representation information and the second result event representation information corresponding to each explicit and implicit causal event. Then, the first causal event representation information and the second causal event representation information are fused to generate the target causal event text data. Finally, the first result event representation information and the second result event representation information are fused to generate the target result event text data. The process of extracting the core components representing the event information uses semantic role labeling and dependency parsing methods.

[0070] In this embodiment, the explicit event text data and the explicit and implicit event text data are parsed and the events are fused to determine multiple target event text data.

[0071] Step 104: Using graph database visualization software, construct a network of cascading complex disaster events related to flooding in the target city based on the text data of each target event.

[0072] Graph database visualization software includes graph databases and complex network analysis software. The graph database is Neo4j, a high-performance open-source graph database based on Java, which is one of the mainstream graph databases. It can effectively organize, store, and update data, and supports functions such as querying and backup. The complex network analysis software is Gephi, an open-source software for network analysis and visualization that provides rich visualization functions and can clearly present network structures.

[0073] It should be noted that in the target city flood cascade complex disaster event network constructed in this invention, nodes represent disaster events, and directed edges (causal chains) represent the interaction relationships between disasters. The process of constructing the target city flood cascade complex disaster event network includes the three steps mentioned above: event causal relationship extraction, event representation, event fusion, and event storage and visualization in this step.

[0074] Furthermore, the text data of each target cause event and each target result event are used as disaster event nodes, and the explicit and implicit causal relationships and explicit causal relationships are used as directed edges. These are then imported into a graph database to construct an initial urban flood cascade complex disaster event network. Complex network analysis software is used to visualize the initial urban flood cascade complex disaster event network to generate the target urban flood cascade complex disaster event network.

[0075] In this embodiment, graph database visualization software is used to construct a network of cascading complex disaster events of flooding in the target city based on the text data of each target event.

[0076] Step 105: Use a pre-set risk assessment function to assess the target city's flood cascade complex disaster event network and generate urban flood disaster assessment data.

[0077] The urban flood disaster assessment data includes the values ​​of degree centrality, betweenness centrality, proximity centrality, eigenvector centrality, causal chain (directed edge) betweenness centrality, causal chain vulnerability, and the global efficiency coefficient of the target city's flood cascade complex disaster event network.

[0078] The pre-defined risk assessment functions include degree centrality calculation function, betweenness centrality calculation function, proximity centrality calculation function, eigenvector centrality calculation function, causal chain (directed edge) betweenness centrality calculation function, causal chain vulnerability calculation function, and global efficiency coefficient calculation function.

[0079] It should be noted that the disaster event node data and directed edge data are extracted from the target city's flood cascade complex disaster event network; the disaster event node data and directed edge data are substituted into a pre-set risk assessment function to calculate the city's flood disaster assessment data; wherein, the disaster event node data includes the in-degree of the disaster event node, the out-degree of the disaster event node, the number of disaster event nodes in the target city's flood cascade complex disaster event network, the set of disaster event nodes in the target city's flood cascade complex disaster event network, the number of nodes that disaster event node j can reach in the target city's flood cascade complex disaster event network, the number of nodes that disaster event node i can reach in the target city's flood cascade complex disaster event network, the number of nodes that disaster event node i can reach in the target city's flood cascade complex disaster event network, the number of connectable disaster event nodes in the target city's flood cascade complex disaster event network, the number of shortest paths between disaster event node s and disaster event node t that pass through disaster event node i, and the number of shortest paths between disaster event node s and disaster event node t.

[0080] The directed edge data includes the shortest path length from disaster event node j to disaster event node i in the target city flood cascade complex disaster event network, the shortest path length from disaster event node i to disaster event node j in the target city flood cascade complex disaster event network, the adjacency matrix of the target city flood cascade complex disaster event network, the number of shortest paths between disaster event node s and disaster event node t that pass through causal chain e, the number of connectable disaster event nodes in the target city flood cascade complex disaster event network after removing causal chain e, and the shortest path length between disaster event node s and disaster event node t.

[0081] Furthermore, the steps described above for substituting disaster event node data and directed edge data into a pre-defined risk assessment function to calculate urban flood disaster assessment data are as follows: Substituting the in-degree and out-degree of disaster event nodes into the degree centrality calculation function to calculate degree centrality; where the in-degree of a disaster event node is the number of points from other disaster event nodes to a given disaster event node, and the out-degree of a disaster event node is the number of points from a given disaster event node to other disaster event nodes; degree centrality is the most direct measure of node centrality in network analysis. It comprehensively considers the in-degree and out-degree of a node, reflecting the degree of association of a node in the network. The higher the degree centrality of a node, the higher its association with other nodes, the more important it is in the network, and the more important and difficult to control the corresponding disaster event is in the disaster chain. The degree centrality calculation function is as follows:

[0082]

[0083] Among them, C D(i) represents degree centrality; N is the number of disaster event nodes in the target city's flood cascade complex disaster event network; d in (i) represents the in-degree of the disaster event node; d out (i) represents the out-degree of the disaster event node.

[0084] Furthermore, the set of disaster event nodes, the number of shortest paths between disaster event nodes s and t passing through disaster event node i, and the number of shortest paths between disaster event nodes s and t in the target city's cascading complex disaster event network are substituted into the betweenness centrality calculation function to calculate betweenness centrality. Betweenness centrality is an indicator that measures the control and constraint effect of a node (or edge) on other nodes in the network. It reflects the number of times a node (or edge) acts as a bridge between two other nodes on the shortest path. The higher the betweenness centrality, the stronger the control and constraint effect of the node (or edge) on the network, and the more significant its bridging and hub role in the network. For disaster event networks, if effective control of the event node or event causal chain can be achieved, the connection between disaster events can be quickly suppressed or even blocked. The betweenness centrality calculation function is as follows:

[0085]

[0086] Among them, C B (i) represents the centrality of the intermediaries; N represents the number of disaster event nodes in the target city's flood cascade complex disaster event network; n(s,t|i) represents the number of shortest paths between disaster event node s and disaster event node t that pass through disaster event node i; I represents the set of disaster event nodes in the target city's flood cascade complex disaster event network; n(s,t) represents the number of shortest paths between disaster event node s and disaster event node t.

[0087] Furthermore, the number of nodes reachable from disaster event node i by disaster event node j in the target city flood cascade complex disaster event network, the number of nodes reachable from disaster event node i by disaster event node j in the target city flood cascade complex disaster event network, the shortest path length from disaster event node j to disaster event node i in the target city flood cascade complex disaster event network, and the shortest path length from disaster event node i to disaster event node j in the target city flood cascade complex disaster event network are substituted into the proximity centrality calculation function to calculate the proximity centrality. Here, proximity centrality reflects the degree of proximity between a node and other nodes, and is an indicator of a node's ability to influence other nodes through the network. The higher the proximity centrality of a node, the smaller the average shortest distance from that node to other nodes, and the more central it is in the network. The proximity centrality of a node is the reciprocal of the average shortest distance from all reachable nodes to that node. Similar to degree centrality, in directed networks, both the direction of the node and the direction of origin from the node need to be considered simultaneously. The proximity centrality calculation function is as follows:

[0088]

[0089] Among them, C c (i) indicates proximity centrality; N in (i) represents the number of nodes that disaster event node j can reach in the target city's flood cascade complex disaster event network; L(j,i) represents the shortest path length from disaster event node j to disaster event node i in the target city's flood cascade complex disaster event network; N out (i) represents the number of nodes in the target city flood cascade complex disaster event network that disaster event node i can reach; L(i,j) represents the shortest path length from disaster event node i to disaster event node j in the target city flood cascade complex disaster event network.

[0090] Furthermore, the adjacency matrix of the target city's flood cascade complex disaster event network is substituted into the eigenvector centrality calculation function to calculate the eigenvector centrality. Eigenvector centrality considers both the importance of a node itself and its neighboring nodes. Its basic idea is that a node's centrality is related to its neighboring nodes; the more important the neighboring nodes, the more important the node is in the network. A node's eigenvector centrality is proportional to the sum of the centrality scores of its neighboring nodes. The larger the eigenvector centrality of a node, the more important its neighboring nodes are. In a disaster event network, the eigenvector centrality of a disaster event characterizes the importance of that event and its neighboring events. Eigenvector centrality is essentially the eigenvector of the network's adjacency matrix; the eigenvector centrality calculation function is as follows:

[0091]

[0092] Where A is the adjacency matrix; x is the eigenvector of the adjacency matrix; λ is the eigenvalue of the adjacency matrix; C E (i) represents the eigenvector centrality; x i Let x1 be the eigenvector centrality value of the i-th disaster event node; x2 be the eigenvector centrality value of the 1st disaster event node; x3 be the eigenvector centrality value of the 2nd disaster event node; x4 be the eigenvector centrality value of the 3rd disaster event node; x5 be the eigenvector centrality value of the 3rd disaster event node; x6 be the eigenvector centrality value of the 4th disaster event node; x7 be the ei N is the value of the eigenvector centrality of the Nth disaster event node; T is the transpose.

[0093] Furthermore, the number of shortest paths passing through causal chain e in the shortest path between disaster event node s and disaster event node t, and the number of shortest paths between disaster event node s and disaster event node t, are substituted into the causal chain betweenness centrality calculation function to calculate the causal chain betweenness centrality. This invention simultaneously considers two indicators—edge betweenness centrality and network connectivity—to evaluate the causal chain of the disaster event network. The concept of edge betweenness centrality is similar to that of node betweenness centrality, reflecting the role of edges as network bridges. The larger the betweenness centrality of an edge, the greater its vulnerability. The causal chain betweenness centrality calculation function is as follows:

[0094]

[0095] Among them, EC B (e) represents the causal chain mediation centrality; N represents the number of disaster event nodes in the target city flood cascade complex disaster event network; n(s,t|e) represents the number of shortest paths between disaster event node s and disaster event node t that pass through causal chain e; n(s,t) represents the number of shortest paths between disaster event node s and disaster event node t; I represents the set of disaster event nodes in the target city flood cascade complex disaster event network.

[0096] Furthermore, the number of connectable disaster event nodes in the target city's flood cascade complex disaster event network, the number of connectable disaster event nodes in the target city's flood cascade complex disaster event network after removing causal chain e, and the causal chain betweenness centrality are substituted into the causal chain vulnerability calculation function to calculate the causal chain vulnerability value; where connectivity characterizes the degree of connectivity or reachability of various disaster events in the network. If the network connectivity decreases after removing an edge, it indicates that the edge plays an important role in the network's connectivity. Therefore, the smaller the network connectivity after removing an edge from the causal chain, the greater the vulnerability of the edge. This invention uses the number of connected edges in the network to characterize the network connectivity. Taking into account both the edge betweenness centrality and connectivity, the causal chain vulnerability calculation function is as follows:

[0097]

[0098] Where EV(e) is the value of causal chain fragility; EC B (e) represents the mediating centrality of the causal chain; denoted as the maximum value of the intermediary centrality of all causal chains in the target city's flood cascade complex disaster event network; ECN is the number of connectable disaster event nodes in the target city's flood cascade complex disaster event network; ECN(e) is the number of connectable disaster event nodes in the target city's flood cascade complex disaster event network after removing causal chain e.

[0099] Furthermore, the shortest path length between disaster event node s and disaster event node t is substituted into the global efficiency coefficient calculation function to calculate the global efficiency coefficient of the target city's flood cascade complex disaster event network. In a directed unweighted network, the shortest path length refers to the number of directed edges on the shortest path from one node to another. Clearly, the network's propagation efficiency is positively correlated with the reciprocal of the shortest path length. The global efficiency coefficient obtained from the global efficiency coefficient calculation function can assess the propagation efficiency of disaster events in the network and also analyze the effectiveness of the evaluation method. The global efficiency coefficient calculation function is as follows:

[0100]

[0101] Where GE is the global efficiency coefficient of the target city flood cascade complex disaster event network; N is the number of disaster event nodes in the target city flood cascade complex disaster event network; I is the set of disaster event nodes in the target city flood cascade complex disaster event network; and L(s,t) is the shortest path length between disaster event node s and disaster event node t.

[0102] In this embodiment, a pre-set risk assessment function is used to assess the network of cascading complex flood disaster events in the target city, generating urban flood disaster assessment data.

[0103] In this embodiment of the invention, a method for constructing and evaluating a network of cascading and complex urban flood disaster events is provided. First, multiple initial urban flood disaster text data are acquired and preprocessed to determine multiple target urban flood disaster text data. Then, a regular expression method and a pre-defined causal extraction model are used to extract causal relationships from the target urban flood disaster text data, generating multiple explicit event text data and multiple explicit and implicit event text data. The explicit and implicit event text data are analyzed and fused to determine multiple target event text data. A graph database visualization software is used to construct a network of cascading and complex urban flood disaster events based on the target event text data. Finally, a pre-defined risk assessment function is used to evaluate the network of cascading and complex urban flood disaster events. The assessment process generates urban flood disaster assessment data. Based on the above scheme, multiple explicit and implicit event text data are obtained using regular expression methods and a pre-set causal extraction model. These explicit and implicit event text data are analyzed and fused. Using graph database visualization software, the obtained target event text data are used to construct a target city flood cascade complex disaster event network. This process considers the interaction between flood-induced disaster events, which helps reveal the development patterns of urban flood cascade complex disaster events. Furthermore, by using a pre-set risk assessment function to evaluate the target city flood cascade complex disaster event network and generate urban flood disaster assessment data, a comprehensive and effective assessment of the disaster event network can be achieved, further reducing the risks posed by urban flood disasters.

[0104] Please see Figure 3 , Figure 3 A flowchart illustrating the steps of another method for constructing and evaluating a network of cascading and complex urban flood disaster events provided in this embodiment of the invention.

[0105] This invention provides a method for constructing and evaluating a network of cascading and complex urban flood disaster events, comprising:

[0106] Step 301: Obtain multiple initial urban flood disaster text data, preprocess each initial urban flood disaster text data, and determine multiple target urban flood disaster text data.

[0107] Further, step 301 may include the following sub-steps:

[0108] S11. Deduplicate the initial urban flood disaster text data to generate deduplicated urban flood disaster text data;

[0109] It should be noted that since the data is obtained based on multiple keywords, different keywords may yield the same data. For example, when the text data of a Weibo post contains both the keywords "flood" and "flooding," it will be crawled repeatedly by the program. In this case, it is necessary to delete duplicate data. Specifically, the initial text data on flood disasters in each city is deduplicated to generate deduplicated text data on flood disasters in each city.

[0110] S12. Perform denoising operation on the deduplicated urban flood disaster text data of each city to generate denoised urban flood disaster text data;

[0111] It should be noted that due to the polysemy and ambiguous word segmentation in Chinese, data unrelated to flood disasters may be incorrectly crawled by the program. For example, if a Weibo post contains the term "pulmonary edema," it will be incorrectly crawled. Furthermore, a significant amount of data on Weibo is related to celebrities, which can interfere with the data and needs to be removed. Research into the characteristics of Weibo data reveals that most of this data contains tags such as "Idol Energy Month" or "Idol V Energy." Therefore, removing Weibo data containing these topics can effectively reduce interference. Specifically, denoising operations are performed on the flood disaster text data for each deduplicated city to generate denoised city flood disaster text data.

[0112] S13. Based on the pre-set place name extraction model, latitude and longitude are used to filter the text data of flood disasters in various denoised cities to determine the text data of flood disasters in multiple target cities.

[0113] In this embodiment, multiple initial urban flood disaster text data are acquired, and each initial urban flood disaster text data is preprocessed to determine multiple target urban flood disaster text data.

[0114] Step 302: Use regular expressions and a pre-set causal extraction model to extract causal relationships from the flood disaster text data of each target city, generating multiple explicit event text data and multiple explicit and implicit event text data.

[0115] Explicit event text data includes explicit causal relationships, explicit cause events, and explicit result events.

[0116] Textual data on explicit and implicit events includes explicit and implicit causal relationships, explicit and implicit cause events, and explicit and implicit result events.

[0117] Further, step 302 may include the following sub-steps:

[0118] S21. Use regular expressions to extract explicit causal relationships from the text data of flood disasters in each target city, and output multiple explicit causal relationships, multiple explicit cause events, and multiple explicit result events;

[0119] It should be noted that a regular expression is a template composed of a sequence of characters and special characters, used to describe the text pattern to be matched. A specific template is shown below:

[0120] ① Cause-and-effect front-end approach

[0121] Examples of causal indicator words: in order to, because, due to

[0122] Cause and effect template:<Conj|Prep...> (Cause), (Effect)

[0123] Example 1: (Because of) (river water overflowing), (the entire underground parking garage was almost flooded).

[0124] ② From cause to effect, the central approach

[0125] Examples of causal indicative words: therefore, cause, lead to, result in, bring about, cause, trigger, lest

[0126] Cause and effect template: (Cause)<Conj|Verb|Adverb...> (Effect)

[0127] Example 2: (Localized flooding disaster) <resulting in> (15 deaths and 2 missing persons in a certain province).

[0128] ③ Cause-and-effect front-end matching

[0129] Causal indicator: <affected by, influenced by>

[0130] Cause and effect template: <conj>(Cause) <verb>(Effect)

[0131] Example 3: Due to the impact of heavy rain and flooding, traffic is slow.

[0132] ④ From cause to effect, a centrally integrated approach

[0133] Examples of causal indicators: 〈due to, therefore〉, 〈because, therefore〉

[0134] Cause and effect template: <conj>(Cause), <conj>(Effect)

[0135] Example 4: Due to severe water accumulation, there is a serious traffic jam at this location.

[0136] (2) Tracing the cause from the effect

[0137] ⑤ Tracing the cause from the effect to the middle method

[0138] Examples of cause-and-effect indicators: because, due to

[0139] Cause-and-effect template: (Effect)<Conj|Prep...> (Cause)

[0140] Example 6: This place (is always flooded when it rains) <because> (the Lai Chi Wan project blocked the drainage outlets leading to the river for the entire street).

[0141] ⑥ Cause-and-effect matching method

[0142] Examples of causal indicators: <cause, reason>, <cause, reason>, <[of] therefore, because>

[0143] Cause and effect template: <conj>(Effect), <conj>(Cause)

[0144] Example 5: The main cause of this (casualties) was (subway backflow).

[0145] Furthermore, to avoid the influence of a single causal cue word on a dual causal cue word, this invention sets priorities for the above six rules as shown in Table 2. In addition, considering that some long sentences contain multiple causal relationships, this invention performs secondary extraction of the extracted causal and result events using the ERNIE model, resulting in more comprehensive extracted information.

[0146] Table 2 Causal Template Rule Priority

[0147]

[0148] S22. Using a pre-set causal extraction model, explicit and implicit causal relationships are extracted from the text data of flood disasters in each target city, generating multiple explicit and implicit causal relationships, multiple explicit and implicit cause events, and multiple explicit and implicit result events.

[0149] It should be noted that this invention uses a deep learning method based on the ERNIE model to extract explicit and implicit causal relationships. Three fully connected neural networks are added to the ERNIE model to perform different tasks, enabling the identification of causal subjects, causal predicates, causal conjunctions, result subjects, and result predicates. First, it determines whether the word is located within a causal relationship region; second, it determines whether the word belongs to either cause or effect; finally, it determines whether the word belongs to subject, predicate, or causal conjunction. Through multi-task learning, the identification of causal subjects, causal predicates, causal conjunctions, result subjects, and result predicates is achieved.

[0150] In this embodiment, regular expressions and a pre-set causal extraction model are used to extract causal relationships from the flood disaster text data of each target city, generating multiple explicit event text data and multiple explicit and implicit event text data.

[0151] Step 303: Analyze the text data of each explicit event and the text data of each explicit and implicit event, and perform event fusion to determine multiple target event text data.

[0152] The target event text data includes target cause event text data and target result event text data.

[0153] Furthermore, step 303 may include the following sub-steps:

[0154] S31. Extract the main body of each explicit cause event and each explicit result event to generate multiple first cause event representation information and multiple first result event representation information;

[0155] S32. Extract the main body of each explicit and implicit cause event and each explicit and implicit result event to generate multiple second cause event representation information and multiple second result event representation information;

[0156] It should be noted that the causal and result events (explicit causal and result events, explicit and implicit causal and result events) obtained through regular expression pattern matching and pre-built causal extraction models still contain redundant information, requiring the extraction of the core components to represent event information. Event representation generally uses semantic role labeling and dependency parsing methods. Semantic role labeling focuses on the predicate of the sentence, analyzing the relationship between each component and the predicate, including core semantic roles (such as agent and patient) and auxiliary semantic roles (such as time and place). Dependency parsing mainly analyzes the dependency relationships between sentence components to reveal the syntactic structure, mainly including subject-predicate relations, verb-object relations, attributive-head relations, and verb-complement structures. Therefore, the event representation method of this invention is mainly divided into two parts:

[0157] Step (1) Semantic role labeling: Extract the subject-verb-object triplet of the event directly based on the core semantic role. If there is no complete subject-verb-object structure, the event is represented by the subject-verb and verb-object structures in sequence. If the extraction result is empty, proceed to step (2).

[0158] Step (2) Dependency parsing: Use dependency parsing to extract the subject-verb-object triplet of the event. If there is no complete subject-verb-object structure, extract the subject-verb relation, verb-object relation, verb and noun in sequence until the event is represented.

[0159] S33. Perform event fusion on each first cause event representation information, each second cause event representation information, each first result event representation information, and each second result event representation information to generate target cause event text data and target result event text data.

[0160] It should be noted that the disaster events extracted above (first cause event representation information and second cause event representation information, first result event representation information and second result event representation information) focus on a single case, and therefore need further generalization. For example, the events "multi-section water accumulation" and "road water accumulation" both represent "water accumulation" events. The main task of event fusion is to merge similar events and express them in a general and concise language. To reduce the workload, this invention first uses the K-Means algorithm for text clustering, and then combines it with manual judgment of events with high similarity to achieve event fusion.

[0161] In this embodiment, the explicit event text data and the explicit and implicit event text data are parsed and the events are fused to determine multiple target event text data.

[0162] Step 304: Based on the graph database, construct an initial urban flood cascaded complex disaster event network by using the text data of each target cause event and the text data of each target result event as disaster event nodes, and the explicit and implicit causal relationships and the explicit causal relationships as directed edges.

[0163] It should be noted that the initial urban flood cascading complex disaster event network is structurally a directed cyclic graph, and therefore needs to be stored in a graph database. Neo4j is a high-performance open-source graph database based on Java, and is one of the mainstream graph databases currently available. It can effectively organize, store, and update data, and supports functions such as querying and backup. This invention utilizes Neo4j to store the constructed disaster event network, so that the network can be updated as needed.

[0164] In this embodiment, based on a graph database, the text data of each target cause event and the text data of each target result event are used as disaster event nodes, and each explicit and implicit causal relationship and each explicit causal relationship are used as directed edges to construct an initial urban flood cascaded complex disaster event network.

[0165] Step 305: Use complex network analysis software to visualize the initial urban flood cascade complex disaster event network and generate the target urban flood cascade complex disaster event network.

[0166] It should be noted that, in order to better understand the development process and structural characteristics of disaster event networks, this invention uses Gephi software for network visualization. Gephi is an open-source software for network analysis and visualization. It provides rich visualization capabilities and can clearly present network structures.

[0167] In this embodiment, complex network analysis software is used to visualize the initial urban flood cascade complex disaster event network and generate the target urban flood cascade complex disaster event network.

[0168] Step 306: Use a pre-set risk assessment function to assess the network of cascading complex flood disaster events in the target city and generate urban flood disaster assessment data.

[0169] Furthermore, step 306 may include the following sub-steps:

[0170] S61. Extract disaster event node data and directed edge data from the target city's flood cascade complex disaster event network;

[0171] S62. Substitute the disaster event node data and directed edge data into the preset risk assessment function to calculate urban flood disaster assessment data.

[0172] It should be noted that the disaster events in the target city flood cascade complex disaster event network constructed in this invention are nodes of a complex network, and the causal relationships between disaster events are directed edges of the complex network. The target city flood cascade complex disaster event network constructed in the above steps of this invention uses complex network theory to identify important disaster events and vulnerable causal chains in the network, and quantifies their importance and vulnerability. First, the importance of key event nodes in the network is evaluated by calculating the degree centrality, betweenness centrality, proximity centrality, and eigenvector centrality of nodes. Second, the vulnerability of disaster causal chains is evaluated by comprehensively considering the betweenness centrality of edges and network connectivity. Finally, the global efficiency coefficient is used to evaluate the global transmission efficiency of the network before and after removing important disaster event nodes and vulnerable causal chains to analyze the validity of the evaluation results.

[0173] In this embodiment, a pre-set risk assessment function is used to assess the network of cascading complex flood disaster events in the target city, generating urban flood disaster assessment data.

[0174] For example, please refer to Figure 4 First, initial urban flood disaster text data was obtained from the Sina Weibo platform using Python web crawler technology based on XPath path language. The initial urban flood disaster text data was deduplicated to generate deduplicated urban flood disaster text data. Denoising was then performed on the deduplicated urban flood disaster text data to generate denoised urban flood disaster text data (highly correlated flood disaster data). Based on a pre-set place name extraction model, the latitude and longitude of the denoised urban flood disaster text data was filtered to determine multiple target urban flood disaster text data. Event causality extraction, event representation, and event... The four processing steps of fusion, event storage, and visualization are as follows: parsing and fusing explicit and implicit event text data to identify multiple target event text data; using graph database visualization software to construct a target city flood cascade complex disaster event network based on the target event text data; and finally, calculating the values ​​of degree centrality, betweenness centrality, proximity centrality, eigenvector centrality, causal chain (directed edge) betweenness centrality, causal chain vulnerability, and the global efficiency coefficient of the target city flood cascade complex disaster event network through a pre-set risk assessment function to complete the network evaluation.

[0175] As a comparison of technological effects, existing technologies can be used as a reference. Current urban flood network construction and risk assessment technologies primarily study single disaster events, neglecting the inherent logical connections and relationships between disasters, and ignoring potential cascading and amplification effects. However, disaster events are usually not independent or unrelated, especially large-scale natural disasters. Studying only a single disaster event is insufficient to reveal its underlying mechanisms. Therefore, with a deeper understanding, scholars have gradually begun to study multiple disasters. Foreign scholar Menoni S., using the 1995 Kobe earthquake as an example, analyzed the interactions between secondary and derivative disaster events, and first proposed the concept of "disaster loss and failure chain." Domestic scholar Guo Zengjian proposed the concept of "disaster chain" in 1987, initially classifying disaster chains into four types: causal chain, homologous chain, mutually exclusive chain, and evenly matched chain. Since then, scholars have applied disaster chain theory in various research fields. KeS-S et al. constructed a disaster chain assessment to evaluate the probability and risk of traffic system congestion caused by building collapse during earthquakes, providing a reference for local governments in emergency management during earthquake disasters; Liu Jiahong et al. conducted research on the "external flood-internal waterlogging" disaster chain, revealing its formation and evolution patterns, and summarizing key scientific and technological issues in the joint prevention and control of external flood and internal waterlogging disaster chains in megacities. Although some progress has been made in disaster chain research both domestically and internationally, research on network risk assessment of cascading complex disaster events related to urban flooding is scarce.

[0176] Furthermore, the challenge of applying complex network theory to urban flood disasters lies in constructing a network of cascading and complex urban flood disaster events. This is because official disaster data (disaster investigation reports, infrastructure data, etc.) is often difficult to obtain, has a time lag, and generally cannot encompass the entire disaster process. The accessibility, comprehensiveness, and real-time nature of social media data can effectively address these issues. Sina Weibo is one of the most widely used social media platforms in my country, boasting numerous official and news media accounts as well as a large number of ordinary users. When floods occur, the high level of public attention to flood disasters leads to a surge in flood-related information on the Weibo platform, providing a reliable data source for the construction and real-time updating of a network of cascading and complex urban flood disaster events.

[0177] Furthermore, the concept of event-based knowledge graphs was proposed by Professor Liu Ting at the 2017 China Computing Conference, aiming to construct an event logic knowledge base to reveal the evolutionary patterns of events and the inherent logical connections between them. Similar research has preceded this; for example, Chambers N et al. automated statistical script extraction and designed a "word-filling" method to evaluate script event prediction models. Zheng Xin et al. used multiple constraints such as reflexivity, transitivity, and event type pairs to propose a Chinese event temporal relationship reasoning method based on global optimization. Currently, event-based knowledge graphs are mainly applied in fields such as finance, consumer affairs, online public opinion, and emergencies. Liu Zhenghao et al. selected high-quality financial emergency text data from financial news reports to construct a two-layer event-based knowledge graph, revealing the evolutionary logic and risk transmission patterns of financial emergencies. Ju Hailong and Peng Jun used a consumer intelligence acquisition method based on causal event-based knowledge graphs, combining rule and dependency parsing analysis to extract event causal relationships and event knowledge, constructing a causal event-based knowledge graph of user purchasing behavior, providing a reference for expanding big data intelligence mining methods. Event graphs have also found applications in urban flooding disaster management. For example, Feng Jun et al. constructed an urban flooding event graph using a deep neural network fusion method based on a voting mechanism. They then used the event graph to generate and train a discrete dynamic Bayesian network, and conducted causal analysis of flooding points on this network. Since its inception, event graphs have been explored by many researchers. Event graphs reveal the evolutionary patterns of events and the inherent logical connections between them, making them highly promising for disaster emergency response and risk assessment. However, few researchers currently focus on the application of event graphs in cascading urban flooding disasters. Existing urban flooding risk assessment techniques primarily assess the severity of flooding disasters and the resulting losses and impacts, neglecting the interactions between flood-related disaster events. Furthermore, current disaster data acquisition is difficult and often outdated, lacking detailed information on the development process of flooding disasters, making it difficult to quickly construct a comprehensive network of urban flooding cascading disaster events.

[0178] To address the aforementioned problems, this invention proposes a method for constructing and evaluating urban flood cascading complex disaster event networks. Utilizing textual data from social media, characterized by its unstructured nature, low quality density, and large volume, this method extracts effective information from social media data to construct urban flood cascading complex disaster event networks. It not only rapidly extracts effective information from massive amounts of social media data for constructing these networks but also applies complex network theory to characterize the development process of urban flooding and identifies and evaluates important disaster events and vulnerable causal chains. This reveals the event evolution patterns of flood cascading complex disasters and the inherent logical connections between events, thereby rapidly identifying and evaluating important disaster events and vulnerable causal chains within the urban flood cascading complex disaster network. Furthermore, it considers… This study considers the interactions between disaster events, which helps to reveal the development patterns of cascading and complex urban flood disasters, providing a theoretical basis for chain-breaking disaster mitigation. It addresses the technical problem that "existing urban flood risk assessment technologies mainly assess the severity of flood disasters and the losses and impacts they cause, while neglecting the interactions between flood-related disaster events." Furthermore, it analyzes flood disaster risks from a network perspective, considering the cascading, superposition, and methodological effects caused by the interactions of disaster events. This enables the identification and assessment of key disaster events and causal chains in the disaster event network. Based on event graphs, it can efficiently extract effective information from massive amounts of unstructured, low-quality, and low-density social media data, thus constructing a more comprehensive network of cascading and complex urban flood disaster events. Simultaneously, due to the availability and real-time nature of social media data, the network can be updated in real time as needed, enabling rapid construction and risk assessment of the disaster event network.

[0179] Furthermore, this invention constructs a cascading complex disaster event network for urban flooding based on event graphs and using social media data as the data source. Using social media data as the data source allows for real-time data updates as needed, helping the public and government departments clearly understand important disaster events and vulnerable causal chains in the development process of floods. This facilitates scientific disaster avoidance, reduces the risks posed by urban floods, and enables relatively comprehensive acquisition of detailed information on the development process of floods. It achieves rapid real-time updates of the disaster event network, solving the technical problems of "currently, disaster data acquisition is difficult, has a lag, and lacks detailed information on the development process of floods." Simultaneously, by utilizing text content from the Sina Weibo platform, a cascading complex disaster event network for urban flooding is constructed based on event graphs to represent the development process of urban floods. Complex network theory is applied to identify important disaster events and vulnerable causal chains within the cascading complex disaster event network, thereby achieving rapid assessment of the urban flooding cascading complex disaster event network.

[0180] In summary, this invention proposes a method for constructing and assessing urban flood cascading complex disaster event networks. It introduces the concept of "urban flood cascading complex disaster events" and presents a method for assessing the network risk of urban flood cascading complex disaster events using social media data as the data source, based on event graphs and applying complex network theory. This method, from a network perspective, can analyze the inherent logical relationships within the disaster event network and identify important disaster events and vulnerable causal chains within the network. Simultaneously, this method can acquire data and update the disaster event network in real time as needed, helping governments and the public to promptly break chains and mitigate disasters during urban flooding, preventing further expansion of disaster impacts due to cascading effects. Furthermore, due to the availability and low cost of data, and its applicability to assessing the network risk of flood cascading complex disaster events in different cities, this method has broad applicability and significant application potential.

[0181] In this embodiment of the invention, a method for constructing and evaluating a network of cascading and complex urban flood disaster events is provided. First, multiple initial urban flood disaster text data are acquired and preprocessed to determine multiple target urban flood disaster text data. Then, a regular expression method and a pre-defined causal extraction model are used to extract causal relationships from the target urban flood disaster text data, generating multiple explicit event text data and multiple explicit and implicit event text data. The explicit and implicit event text data are analyzed and fused to determine multiple target event text data. A graph database visualization software is used to construct a network of cascading and complex urban flood disaster events based on the target event text data. Finally, a pre-defined risk assessment function is used to evaluate the network of cascading and complex urban flood disaster events. The assessment process generates urban flood disaster assessment data. Based on the above scheme, multiple explicit and implicit event text data are obtained using regular expression methods and a pre-set causal extraction model. These explicit and implicit event text data are analyzed and fused. Using graph database visualization software, the obtained target event text data are used to construct a target city flood cascade complex disaster event network. This process considers the interaction between flood-induced disaster events, which helps reveal the development patterns of urban flood cascade complex disaster events. Furthermore, by using a pre-set risk assessment function to evaluate the target city flood cascade complex disaster event network and generate urban flood disaster assessment data, a comprehensive and effective assessment of the disaster event network can be achieved, further reducing the risks posed by urban flood disasters.

[0182] Please see Figure 5 , Figure 5 This is a structural block diagram of a device for constructing and evaluating a network of cascading complex disaster events in urban flooding, provided as an embodiment of the present invention.

[0183] This invention provides a device for constructing and evaluating a network of cascading and complex urban flood disaster events, comprising:

[0184] The acquisition module 501 is used to acquire multiple initial urban flood disaster text data, preprocess each initial urban flood disaster text data, and determine multiple target urban flood disaster text data;

[0185] Module 502 is used to extract causal relationships from flood disaster text data of various target cities using regular expressions and a pre-built causal extraction model, generating multiple explicit event text data and multiple explicit and implicit event text data.

[0186] The parsing module 503 is used to parse the text data of each explicit event and the text data of each explicit and implicit event, and to perform event fusion to determine multiple target event text data;

[0187] Module 504 is used to construct a network of cascading complex flood disaster events in a target city based on textual data of each target event using graph database visualization software.

[0188] The generation module 505 is used to evaluate the network of cascading complex disaster events of floods in the target city using a pre-set risk assessment function, and generate urban flood disaster assessment data.

[0189] Furthermore, module 501 includes:

[0190] The deduplication submodule is used to deduplicatize the initial urban flood disaster text data and generate deduplicated urban flood disaster text data.

[0191] The denoising submodule is used to perform denoising operations on the deduplicated urban flood disaster text data and generate denoised urban flood disaster text data.

[0192] The filtering submodule is used to filter the text data of flood disasters in various denoised cities by latitude and longitude based on a pre-set place name extraction model, and to determine the text data of flood disasters in multiple target cities.

[0193] Furthermore, explicit event text data includes explicit causal relationships, explicit cause events, and explicit result events; explicit and implicit event text data includes explicit and implicit causal relationships, explicit and implicit cause events, and explicit and implicit result events; module 502 is used, including:

[0194] The explicit extraction submodule is used to extract explicit causal relationships from the text data of flood disasters in each target city using regular expressions, and outputs multiple explicit causal relationships, multiple explicit cause events, and multiple explicit result events.

[0195] The implicit extraction submodule is used to extract explicit and implicit causal relationships from the text data of flood disasters in each target city using a pre-built causal extraction model, generating multiple explicit and implicit causal relationships, multiple explicit and implicit cause events, and multiple explicit and implicit result events.

[0196] Furthermore, the target event text data includes target cause event text data and target result event text data; the parsing module 503 includes:

[0197] The first main branch extraction submodule is used to extract the main branch of each explicit cause event and each explicit result event, and generate multiple first cause event representation information and multiple first result event representation information.

[0198] The second backbone extraction submodule is used to extract the backbone of each explicit and implicit cause event and each explicit and implicit result event, and generate multiple second cause event representation information and multiple second result event representation information.

[0199] The fusion submodule is used to fuse the first cause event representation information, the second cause event representation information, the first result event representation information, and the second result event representation information respectively, to generate target cause event text data and target result event text data.

[0200] Furthermore, the graph database visualization software includes graph databases and complex network analysis software, and module 504 includes:

[0201] The initial construction submodule is used to construct an initial urban flood cascaded complex disaster event network based on a graph database, using the text data of each target cause event and the text data of each target result event as disaster event nodes, and each explicit and implicit causal relationship and each explicit causal relationship as directed edges.

[0202] The visualization submodule is used to visualize the initial urban flood cascade complex disaster event network using complex network analysis software, and generate the target urban flood cascade complex disaster event network.

[0203] Furthermore, the generation module 505 includes:

[0204] The extraction submodule is used to extract disaster event node data and directed edge data from the target city's flood cascade complex disaster event network.

[0205] The calculation and assessment data submodule is used to input disaster event node data and directed edge data into a preset risk assessment function to calculate urban flood disaster assessment data.

[0206] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, modules, and sub-modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0207] This invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program; when the computer program is executed by the processor, the processor performs the steps of the urban flooding cascaded complex disaster event network construction and evaluation method as described in any of the above embodiments.

[0208] This invention also provides a computer-readable storage medium storing a computer program / instructions thereon, which, when executed by a processor, implements the steps of the urban flooding cascaded complex disaster event network construction and evaluation method as described in any of the above embodiments.

[0209] This invention also provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the urban flooding cascaded complex disaster event network construction and evaluation method as described in any of the above embodiments.

[0210] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0211] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0212] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.< / conj> < / conj> < / conj> < / conj> < / verb> < / conj>

Claims

1. A method for constructing and evaluating a network of cascading and complex urban flood disaster events, characterized in that, include: Multiple initial urban flood disaster text data are obtained, and each initial urban flood disaster text data is preprocessed to determine multiple target urban flood disaster text data; The causal relationship was extracted from the flood disaster text data of each target city using regular expression method and pre-set causal extraction model, generating multiple explicit event text data and multiple explicit and implicit event text data; The explicit event text data and the explicit and implicit event text data are analyzed and fused to determine multiple target event text data. A network of cascading complex flood disaster events in the target city was constructed using graph database visualization software based on the text data of each target event. A pre-defined risk assessment function is used to assess the network of cascading and complex flood disaster events in the target city, generating urban flood disaster assessment data. The explicit event text data includes explicit causal relationships, explicit cause events, and explicit result events; The explicit and implicit event text data includes explicit and implicit causal relationships, explicit and implicit cause events, and explicit and implicit result events; The step of extracting causal relationships from the flood disaster text data of each target city using regular expressions and a pre-set causal extraction model to generate multiple explicit event text data and multiple explicit and implicit event text data includes: The regular expression method is used to extract explicit causal relationships from the text data of flood disasters in the target cities, and outputs multiple explicit causal relationships, multiple explicit cause events, and multiple explicit result events; A pre-defined causal extraction model is used to extract explicit and implicit causal relationships from the textual data of flood disasters in the target cities, generating multiple explicit and implicit causal relationships, multiple explicit and implicit cause events, and multiple explicit and implicit result events.

2. The method for constructing and evaluating a network of cascading and complex urban flood disaster events according to claim 1, characterized in that, The step of preprocessing the initial urban flood disaster text data to determine multiple target urban flood disaster text data includes: The initial urban flood disaster text data is deduplicated to generate deduplicated urban flood disaster text data; Denoising operations are performed on the deduplicated urban flood disaster text data to generate denoised urban flood disaster text data; Based on a pre-set place name extraction model, latitude and longitude filtering is performed on the denoised urban flood disaster text data to determine multiple target city flood disaster text data.

3. The method for constructing and evaluating a network of cascading and complex urban flood disaster events according to claim 1, characterized in that, The target event text data includes target cause event text data and target result event text data; The step of parsing the explicit event text data and the implicit event text data and performing event fusion to determine multiple target event text data includes: The main body of each explicit cause event and each explicit result event is extracted to generate multiple first cause event representation information and multiple first result event representation information; The main body of each explicit and implicit cause event and each explicit and implicit result event is extracted to generate multiple second cause event representation information and multiple second result event representation information; Event fusion is performed on each of the first cause event representation information, each of the second cause event representation information, each of the first result event representation information, and each of the second result event representation information to generate target cause event text data and target result event text data.

4. The method for constructing and evaluating a network of cascading and complex urban flood disaster events according to claim 3, characterized in that, The graph database visualization software includes graph databases and complex network analysis software; The step of constructing a network of cascading complex flood disaster events in a target city using graph database visualization software based on the text data of each target event includes: Based on a graph database, the text data of each target cause event and the text data of each target result event are used as disaster event nodes, and the explicit and implicit causal relationships and the explicit causal relationships are used as directed edges to construct an initial urban flood cascaded complex disaster event network. The initial urban flood cascade complex disaster event network was visualized using complex network analysis software to generate the target city flood cascade complex disaster event network.

5. The method for constructing and evaluating a network of cascading and complex urban flood disaster events according to claim 4, characterized in that, The step of evaluating the target city's cascading complex disaster event network using a pre-set risk assessment function to generate urban flood disaster assessment data includes: Extract disaster event node data and directed edge data from the target city's flood cascade complex disaster event network; Substitute the disaster event node data and the directed edge data into a preset risk assessment function to calculate urban flood disaster assessment data.

6. A device for constructing and evaluating a network of cascading and complex urban flood disaster events, applied to the method for constructing and evaluating a network of cascading and complex urban flood disaster events as described in claim 1, characterized in that, include: The acquisition module is used to acquire multiple initial urban flood disaster text data, and preprocess each of the initial urban flood disaster text data to determine multiple target urban flood disaster text data; A module is used to extract causal relationships from the flood disaster text data of each target city using regular expressions and a pre-built causal extraction model, generating multiple explicit event text data and multiple explicit and implicit event text data. The parsing module is used to parse the explicit event text data and the explicit and implicit event text data and perform event fusion to determine multiple target event text data; The construction module is used to construct a network of cascading complex flood disaster events in the target city based on the text data of each target event using graph database visualization software; The generation module is used to evaluate the target city's flood cascade complex disaster event network using a pre-set risk assessment function, and generate urban flood disaster assessment data.

7. An electronic device, characterized in that, The device includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method for constructing and evaluating a network of cascading and complex urban flood disaster events as described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method for constructing and evaluating urban flood cascaded complex disaster event networks as described in any one of claims 1-5.

9. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method for constructing and evaluating urban flood cascaded complex disaster event networks as described in any one of claims 1-5.

Images