Post-earthquake disaster emergency dynamic monitoring method based on network big data

Abstract

This invention discloses a method for dynamic monitoring of post-earthquake disaster emergency response based on network big data, comprising: acquiring post-earthquake network data; establishing a structured dataset; performing data analysis on the structured dataset to determine the optimal network earthquake emergency monitoring indicators, and marking unidentifiable and identifiable indicators; decomposing into sub-indicator datasets; extracting information from the sub-indicator datasets; establishing a cloud-based multi-source geographic information database; establishing earthquake emergency monitoring vector point files; obtaining a vector layer with geographic coordinates for the earthquake emergency monitoring indicators; outputting a spatial distribution map of earthquake emergency monitoring; realizing rapid and automatic acquisition, information identification, and location of post-earthquake disaster data; fusing multi-source data; establishing a spatiotemporal visualization image of earthquake emergency data; and realizing dynamic monitoring of earthquake emergency response based on network information.

Description

Technical Field

[0001] This invention relates to the field of earthquake disaster emergency monitoring technology, specifically to a technology that uses artificial intelligence to quickly acquire network earthquake emergency monitoring information, and is a method and system for post-earthquake disaster network big data emergency dynamic monitoring. Background Technology

[0002] Earthquakes are among the most devastating sudden disasters affecting society and the economy. After an earthquake, the rapid and accurate acquisition of disaster data is crucial for providing information support for rescue efforts. Earthquake monitoring data provides a scientific basis for accurately formulating rescue plans and assessing earthquake damage.

[0003] Currently, the conventional method for earthquake disaster emergency monitoring is to collect data through manual on-site investigations and use satellite remote sensing technology to interpret the damage status of physical objects. This method is time-consuming, collects incomplete information, and is labor-intensive, making it difficult to guarantee the needs of rapid earthquake emergency monitoring. Summary of the Invention

[0004] In view of the problems existing in the prior art, this invention provides a method for rapidly acquiring earthquake emergency monitoring information using network big data. Through the technical solution of this invention, rapid and automatic acquisition, information identification, and location of post-earthquake disaster data are achieved; multi-source data are fused to establish a spatiotemporal visualization image of earthquake emergency data, realizing dynamic earthquake emergency monitoring based on network information.

[0005] The technical solution of this invention:

[0006] The method for dynamic monitoring of post-earthquake disaster emergency response based on network big data includes the following steps:

[0007] 1) Obtain post-earthquake network data. Using social media platforms, media websites, and local government websites related to the earthquake as information sources, and employing legitimate web crawling programs, a raw post-earthquake network dataset was established after targeted optimization of earthquake disaster keywords.

[0008] 2) Raw data processing and invalid data removal. Clean the collected raw data, removing inconsistent, incomplete, invalid, redundant, and header files, and establish a standardized post-earthquake disaster data set.

[0009] 3) Convert the dataset from step 2) into a data table in chronological order. Use regular expressions to decompose and extract the data attributes of time, source, title, and location, including features such as publication time, event time, event location, information source, publisher name (nickname), and information source. Add attribute fields to the data table to create a structured dataset.

[0010] 4) Analyze the structured dataset using TF-ID text keyword extraction technology. Targeting earthquake emergency monitoring needs, determine the optimal network emergency monitoring indicators, including nine indicators: death, injury, housing, roads, rescue teams, relief supplies, affected population, secondary geological disasters, and psychological factors. Establish a dictionary set for these earthquake emergency monitoring indicators, and use word frequency statistics to determine the semantics of each indicator, thus expanding the comprehensiveness and accuracy of data retrieval. Indicators with smaller TF-ID values ​​are marked as unrecognizable indicators because the network cannot directly obtain information from the network; the remaining indicators are marked as recognizable indicators. For unrecognizable indicators, proceed to step 8).

[0011] 5) The structured dataset master table in step 3) is decomposed into multiple dimensions by adding labels to the selected features according to the post-earthquake emergency monitoring indicators. Using artificial intelligence technologies such as Naive Bayes algorithm and LSTM algorithm, it is decomposed into 9 sub-indicator datasets that are arranged in chronological order up to the current time point and include emergency indicator keywords.

[0012] 6) Information extraction from sub-indicator datasets: According to the characteristics of the indicators, extract the numerical and textual descriptive data of the corresponding indicators from each sub-indicator dataset, use data transformation and normalization methods to quantify the values, and save the quantification results as a result data table in chronological order. The location name is recorded in detail using four levels of fields: prefecture, county, town, and village.

[0013] 7) Establish a cloud-based geographic information database to integrate social, economic, and natural data sets of the disaster area.

[0014] The database contains multi-source data in GIS format, including resident population density, vector administrative division maps, roads, water systems, DEM, satellite remote sensing images, POI vector point data, statistical yearbooks, and population census data.

[0015] 8) For unidentifiable indicators, a disaster loss model is established using information obtained from the network and cloud databases as parameters to calculate the unidentifiable indicators.

[0016] 9) Integrate the dataset into the geographic information system, and associate the fourth-level place names and POI vector points with the results data table level by level, add the geographic latitude and longitude coordinates of the event, and establish an earthquake emergency monitoring vector point file.

[0017] 9) Associate the values ​​of the nine indicators with the corresponding vector data to obtain a vector layer with geographic coordinates for each indicator.

[0018] 10) Quantify and numerically map the vector data to output a spatial distribution map of earthquake emergency monitoring.

[0019] Compared with the prior art, the beneficial technical effects of the present invention are as follows:

[0020] This invention provides a method for rapidly acquiring earthquake emergency monitoring information using network big data, comprising: acquiring post-earthquake network data; establishing a structured dataset; performing data analysis on the structured dataset to determine the optimal network earthquake emergency monitoring indicators, and marking unidentifiable and identifiable indicators; decomposing into sub-indicator datasets; extracting information from the sub-indicator datasets; establishing a cloud-based multi-source geographic information database; establishing earthquake emergency monitoring vector point files; obtaining a vector layer with geographic coordinates for the earthquake emergency monitoring indicators; and outputting a spatial distribution map of earthquake emergency monitoring. Using the technical solution of this invention, information can be collected quickly, efficiently, and comprehensively, enabling rapid and automatic acquisition, information identification, and location of post-earthquake disaster data. By fusing multi-source data, a spatiotemporal visualization image of earthquake emergency data can be established, achieving dynamic earthquake emergency monitoring based on network information, improving accuracy, and ensuring the needs of rapid earthquake emergency monitoring are met. Attached Figure Description

[0021] Figure 1 The flowchart of an earthquake disaster emergency monitoring method combining artificial intelligence and network big data provided by the present invention is shown.

[0022] Figure 2 This is a structural block diagram of an earthquake disaster emergency monitoring system that combines artificial intelligence and network big data, as implemented in an embodiment of the present invention. Detailed Implementation

[0023] The present invention will be further described below with reference to the accompanying drawings and embodiments, but the scope of the invention is not limited in any way.

[0024] Figure 1 This invention provides a method and process for earthquake disaster emergency monitoring that combines artificial intelligence and network big data. Specifically, it implements an earthquake disaster emergency monitoring system that combines artificial intelligence and network big data, the structure of which is as follows: Figure 2As shown, the system includes: a post-earthquake network dataset establishment module, a data decomposition module, a quantitative indicator module, a multi-source database module, a prediction model establishment module, an emergency monitoring data spatialization module, and an earthquake emergency monitoring data visualization and mapping module. Specifically, the post-earthquake network dataset establishment module establishes an original post-earthquake disaster dataset based on raw data collection; the data decomposition module performs attribute-structured integration of the data, decomposes the collected records in chronological order, extracts each column of fields, and establishes a standardized database; the quantitative indicator module establishes emergency monitoring indicators, obtaining multiple sub-indicator datasets including emergency indicator keywords; the multi-source database module establishes a cloud-based geographic information database integrating social, economic, and natural multi-source data from the disaster area; the prediction model establishment module establishes estimation and prediction models for network-unidentifiable indicators, including a prediction model for the affected population; the emergency monitoring data spatialization module performs spatial statistics on earthquake emergency monitoring vector data indexed by time intervals, constructing earthquake emergency monitoring time series change data; and the earthquake emergency monitoring data visualization and mapping module displays the spatial distribution characteristics of the disaster and further detects the environmental conditions of the disaster through mapping.

[0025] The specific implementation of the method of the present invention includes the following steps:

[0026] S101 Step 1: Raw data collection, establishing the raw post-earthquake disaster dataset; includes the following operations:

[0027] S102 starts the pre-programmed web crawler data collection program, inputs the preferred post-earthquake information keywords, which consist of the epicenter location name, earthquake time, the administrative province, city and county where the epicenter is located, and the names of important areas in this region.

[0028] S103 is a program that collects news, announcements, and social media messages about the post-earthquake disaster from specific target information source websites, establishing raw data on the disaster situation. The specific information sources are authoritative media outlets, government websites, and social media platforms with a large user base and rapid updates. The selected information sources after comparison are: CCTV News Network, Xinhua News Agency, Weibo, WeChat official accounts, and the websites of local governments affected by the disaster.

[0029] The S104 crawler algorithm follows a combination of depth-first and breadth-first search principles to traverse, analyze, and store internet data, collecting as much earthquake-related information as possible. Specifically, it scores the relevance of selected publicly available webpage texts to keywords to determine whether to include them in the database.

[0030] S105 saves the collected text entries in chronological order into the original post-earthquake dataset to create the original post-earthquake disaster dataset.

[0031] S201 Step 2: Data Cleaning and Processing

[0032] The original post-earthquake disaster data was subjected to information redundancy assessment, symbol information processing, and digital information processing, and useless tags, special symbols, and stop words were deleted; duplicate records were identified and removed using a similarity algorithm model.

[0033] Step 3 of S301: Data Structuring. The data processed in S201 is integrated and structured by attributes. The collected records are decomposed according to time sequence, each column of fields is extracted, and a standardized database, EarthquakeDATA1, is established.

[0034] S302 uses regular expression matching and other techniques to match characters in the text of EarthquakeDATA1, decompose features such as publication time, event time, event location, information source, publisher name (nickname), and information source, and stores them in the database.

[0035] Step 4 of S401: Screen earthquake emergency monitoring indicators based on network big data;

[0036] Earthquake emergency monitoring demand indicators are a set of data characterizing earthquake emergency situations, which can be determined based on the characteristics of network data and emergency management needs. Specifically, this invention utilizes TF-ID (term frequency–inverse document frequency) text keyword extraction technology and the Jieba word segmentation model to analyze the EarthquakeDATA1 dataset, establishing earthquake emergency monitoring indicators for network big data through text keywords. The process is as follows: The earthquake emergency monitoring demand indicator set is denoted as [Index1]; the text collected from the network is segmented using Jieba word segmentation technology to establish a vocabulary sample library data1; the term frequency matrix and term frequency matrix vector of the earthquake emergency monitoring demand indicators are calculated; the TF-IDF value is calculated; and the words are processed through term frequency statistics and inverse document frequency statistics, selecting the top 30 words with high TF-IDF values ​​as the hot vocabulary set of network big data earthquake emergency indicators, denoted as [Index2]. The intersection of the two datasets is calculated, i.e., [Index1]∩[Index2], resulting in [Index3], which represents the network emergency monitoring indicators. These indicators include nine categories: deaths, injuries, housing, roads, rescue teams, disaster relief supplies, affected population, secondary geological disasters, and psychological factors. A network identifiability assessment is added: indicators with high word frequency statistics (>0.1) are identifiable and assigned a value of 1; indicators with low word frequency statistics (<0.1) but high inverse document frequency (>10) are assigned a value of 0, indicating they are unidentifiable. Identifiable and unidentifiable indicators are processed separately in subsequent steps.

[0037] S402 establishes a data indicator dictionary, expanding the descriptions and scope of indicators to accurately identify standards for earthquake emergency monitoring data. Based on network big data statistics, the semantics of the nine indicators are expanded to construct conditions and models for judging monitoring indicators. The model for each indicator is as follows: Y1 = X1(i1, i2….i n Y2=X2(j1,j2,…j m ...Y9=X9(q1,q2q) k ).

[0038] In the above formula, Y1~Y9 represent the text category, X1~X9 represent the earthquake network emergency monitoring index function, and iq represents the corresponding extended vocabulary in the index.

[0039] S501 decomposes the structured data master table of S301 into a multidimensional sub-indicator database by adding labels with hit features according to the post-earthquake emergency monitoring indicators. Using artificial intelligence technologies such as the Naive Bayes algorithm and the LSTM algorithm, a dataset of nine sub-indicators, including emergency indicator keywords, arranged chronologically up to the current time point is established. The multidimensional decomposition method includes steps S502 to S503.

[0040] The S502 data analysis process is as follows: The input text records are segmented and processed. The Naive Bayes algorithm is used to perform probability analysis on each word in the news information after a large amount of text annotation in EarthquakeDATA1. The distribution probability of each word in each type of news is obtained. Each index model is selected for fitting and the probability evaluation result of each sentence type is obtained. Based on the probability evaluation result, the optimal index type of this news is determined.

[0041] S503 outputs the judgment result text to the "details" field of the corresponding sub-indicator dataset, adds an indicator name field, and retains the original record structure and attributes, outputting a database EarthquakeDATA2(I1, I2...I9) with 9 sub-indicator attributes such as release time, detailed information, event information, and source attributes.

[0042] S601 monitoring data information extraction. The EarthquakeDATA2 (I1, I2...I9) sub-indicator dataset was normalized using the LSTM (Long Short-Term Memory) algorithm. Post-earthquake text records were further simplified based on the connection patterns of text words to extract key information and numbers, which were then entered into a results table. The results were then iteratively evaluated line by line in the sub-table to create a results file. The information extraction process involves reading one indicator record E from the EarthquakeDATA2 sub-indicator dataset over a specific period. iFor the i-th statement Ci in the read record, the Jieba segmentation technology is used for word segmentation, converting the words into strings x1, x2, ... The strings are then compared with the data dictionary of the corresponding indicators. Statements Ci' containing the data dictionary are output to a description field in the result dataset. Then, the statements saved in the description field are used to extract information for both numeric and descriptive indicators according to different models. The extracted information Di is saved to a target field in the result dataset. Numeric indicators use the isdigit() function to extract values ​​from the numeric set DI in the result dataset's description field. Descriptive indicators extract information from the descriptive set SI in the result dataset's description field, achieved by superimposing n statements containing the indicator dictionary Ci'. The extracted result is Di. The target data SI is then extracted according to the descriptive indicator model, and the extracted data Di is saved in the target field of the result dataset. The specific formula is as follows:

[0043]

[0044] Finally, the results are updated, and the next time-based sub-indicator dataset EarthquakeDATA2 is retrieved, containing the indicator record E. i2 The extracted results are compared with the target field data in the result dataset. If the numerical indicators are equal, they are not included in the result data record; if the values ​​are different, a new record is added to save the data. For descriptive indicators, semantic similarity analysis is performed, and the C values ​​are compared according to the third similarity theorem. i 'With C i2 The method of determining whether the single value of the segmented strings in the statement is the same is used as the criterion. The strings are compared, and if they are the same, the comparison value is set to "true". i =1, otherwise 0. After calculation by the following formula, P≥0.7 indicates that the statements are the same, so they are not included in the result data record; otherwise, a record is added to save the data.

[0045]

[0046] The target results are saved as a chronologically ordered results table, with location names recorded in detail using four levels of fields: prefecture, county, town, and village. Different indicators among the nine indicators use different recognition models.

[0047] S602 Death Toll: This extracts numbers. Based on the death and judgment model, it identifies key statements related to death and searches for the numbers appearing in them. These numbers are then input into the fields of the results database. The system checks if the next record has the same death toll. If they do, it proceeds to the next judgment. If they do not, it records a new value. This process continues until all records are cycled through. The method for extracting the number of injured and the number of deaths is basically the same.

[0048] S603 Other Indicator Extraction: Damaged Buildings: Based on the building damage model, identify key statements describing the building's condition and retain these statements in text format; Damaged Infrastructure: Based on the infrastructure damage model, record statements describing power and communication damage and save them to database fields; Damaged Transportation: Based on the transportation damage model, save statements describing damaged or intact transportation to database fields, and add a location-based model to extract location and street name information mentioned in the statements; Rescue Teams: Based on the rescue team model, extract statements describing rescue teams, save them to database fields, and run a pre-set model to determine the type of rescue team. The system is used to: 1) Classify aids into categories such as medical, fire, military, rescue teams, and professional teams, and add these categories to the database's category field; 2) Rescue supplies: Run the rescue supplies model to extract descriptions of aid supplies and save them to the database's field; 3) Run a pre-set aid type identification model to identify the types of aids provided, such as water, tents, and blankets, and save these to the aid type field; 4) Secondary disasters: Run the secondary disaster model to extract descriptions of secondary disasters and save them to the database's field; 5) Psychological condition: Run the psychological condition model to extract descriptions of the psychological condition of disaster victims and save them to the database's field.

[0049] S701 establishes a cloud platform database of pre-earthquake data. This database, in GIS format, includes multi-source data such as resident population density, vector administrative division maps, roads, waterways, DEMs, satellite remote sensing images, POI vector point data, statistical yearbooks, and census data.

[0050] S702 Disaster-Affected Population Index Calculation. This index is a network-unidentifiable index. The calculation method adds parameters of floating population density and tourist population density to the original disaster-affected population model. Adding these parameters improves the accuracy of disaster-affected population estimation and ensures that the earthquake affects populations with high population density and greater difficulties in disaster relief and resettlement. The revised estimation model is as follows:

[0051]

[0052] In the above formula, POP dis The total number of affected people, n and m are the number of rows and columns of the grid in the affected area, and the row and column values ​​of the affected area are obtained by overlaying the seismic intensity map and the county-level administrative vector map obtained from the network; POPdis is the number of affected people, I i For earthquake intensity, obtained from online information, P a For intensity probability, POP i Population density and POPs calculated using census data for point (i, j) within the disaster area. f To calculate the density of the floating population using census data, POPs tIt is the tourist population density, calculated using tourism information and hotel distribution in the disaster-stricken scenic area.

[0053] The calculated number of affected people in S703 is input into the earthquake disaster emergency monitoring dataset.

[0054] The S801 dataset was integrated into the Geographic Information System. The process was as follows: the fourth-level place names in the results table were associated with the POI vector points level by level; the geographic latitude and longitude coordinates of the events were added; the point vectors were spatially associated with the administrative county polygons to obtain the accurate county to which the event points belonged and fill them into the county name field of the fourth-level place names, thus establishing an earthquake emergency monitoring vector dataset.

[0055] For digital monitoring indicator datasets, the latitude and longitude coordinates of the locations are extracted using a location association method and stored in the database.

[0056] For text-based monitoring indicators, the location names in the text are read, the latitude and longitude coordinates and geographic coordinates of the locations are extracted and converted, and then stored in the database.

[0057] Post-earthquake emergency monitoring data mapping for S901. On the cloud platform, numerical monitoring indicators are displayed using dots, and heat maps show the most severely affected areas. For text-based monitoring indicators describing the event, real-time high-resolution remote sensing images are used as a base map, overlaid with post-earthquake emergency monitoring data to display the spatial distribution characteristics of the disaster and further assess the environmental conditions related to the disaster.

[0058] S1001 Earthquake Emergency Monitoring Time Series Spatialization. Earthquake emergency monitoring vector datasets were input into the county and township administrative polygons, roads, rivers, and DEM layers of the earthquake area in the GIS. Different time length units were used at different stages of the earthquake: 4 hours within 72 hours after the earthquake, 12 hours from 72 hours to 5 days after the earthquake, and 1 day from 6 to 15 days after the earthquake. Spatial statistics were performed on the earthquake emergency monitoring vector data using time intervals as indices to construct earthquake emergency monitoring time series variation data.

[0059] S1101 Cloud-based Earthquake Emergency Monitoring Information Visualization Charts. The attribute data table from the earthquake emergency monitoring time-series change vector data is uploaded to the cloud database. Using the statistical chart function in Excel, time variation curves and bar charts are generated with time as the X-axis and numerical indicators as the Y-axis.

[0060] It should be noted that the purpose of disclosing the embodiments is to help further understand the present invention. However, those skilled in the art will understand that various substitutions and modifications are possible without departing from the scope of the present invention and the appended claims. Therefore, the present invention should not be limited to the content disclosed in the embodiments, and the scope of protection of the present invention is defined by the scope of the claims.

Claims

1. A method for dynamic monitoring of post-earthquake disaster emergency response based on network big data, comprising the following steps: 1) Obtain post-earthquake network data and establish a raw dataset of post-earthquake disaster network data; Using web crawler tools and optimized earthquake disaster keyword filtering, a post-earthquake online raw dataset was established. The optimized keywords consist of: epicenter location name, earthquake time, administrative province, city, and county where the epicenter is located, and names of important regions in the area; 2) Organize the raw data, including removing invalid data; Clean the collected raw data, remove invalid, redundant and header files that are inconsistent and incomplete in the dataset, and establish a standardized dataset of post-earthquake disaster situation; 3) Convert the dataset from step 2) into a data table in chronological order. Use regular expressions to extract the data attribute features of time, source, title, and location, including: publication time, event time, event location, information source, publisher name, and information source. Add attribute fields to the data table to create a structured dataset. 4) Based on the structured dataset, the optimal network earthquake emergency monitoring indicators are determined by selecting indicators based on earthquake emergency monitoring needs. These indicators include death indicators, injury indicators, housing indicators, road indicators, rescue team indicators, rescue material indicators, affected population indicators, geological secondary disaster indicators, and psychological indicators. A dictionary set of earthquake emergency monitoring indicators is established. Word frequency statistics are used to determine the semantics corresponding to the earthquake emergency monitoring indicators, expanding data retrieval. Unidentifiable and identifiable indicators are identified and labeled. For unidentifiable indicators, proceed to step 8). 5) Perform multidimensional decomposition on the structured dataset from step 3) by adding labels to the selected features according to the post-earthquake emergency monitoring indicators, resulting in multiple sub-indicator datasets arranged in chronological order and including keywords of earthquake emergency monitoring indicators; including: First, data analysis is performed: the input text records are segmented and processed. The Naive Bayes algorithm is used to perform probability analysis on each word in the text-annotated news information in the structured dataset of step 3), to obtain the distribution probability of each word in each type of news, select each index model and fit it, and evaluate the probability of each sentence type through probability. Based on the probability evaluation results, the optimal category of the news is determined. Then, the result text of the judgment is used as the "detailed content" field of the corresponding sub-indicator dataset, an indicator name field is added, and the original record structure and attributes are retained, thus obtaining multiple sub-indicator datasets with release time, detailed information, event information, and source attributes. 6) Extract information from the sub-indicator dataset obtained in step 5); Based on the characteristics of the indicators, the numerical and textual descriptive data of the corresponding indicators in each sub-indicator dataset are extracted, quantified, and the quantification results are saved as a result data table in chronological order; the location names adopt four levels of place names: prefecture, county, town, and village. 7) Establish a cloud-based multi-source geographic information database, in which the multi-source data in GIS format includes resident population density, vector administrative division maps, roads, water systems, DEM, satellite remote sensing images, POI vector point data, statistical yearbooks, and population census data; 8) Based on information obtained from the network and multi-source geographic information databases in the cloud, calculate unidentifiable indicators by establishing a disaster loss model; unidentifiable indicators include disaster-affected population indicators; In establishing a disaster loss model for the affected population, parameters for floating population density and tourist population density are added, expressed as follows: In the above formula, POPdis represents the total number of affected people, U and V are the number of rows and columns of the affected area raster, and the row and column values ​​of the affected area are obtained by overlaying the seismic intensity map obtained from the network and the county-level administrative vector map; Ii （u,v） For the seismic intensity of a grid cell; POPi （u,v） Population density (POPf) is the resident population density in the disaster-stricken area calculated using census data. （u,v） Population density; POPt （u,v） Tourist population density; Pa （u,v） For intensity probability; The obtained dataset can be integrated into a geographic information system to enable spatial visualization and mapping output of post-earthquake data. Through the above steps, we can achieve dynamic monitoring of post-earthquake disaster emergency response based on network big data.

2. The method for dynamic monitoring of post-earthquake disaster emergency response based on network big data as described in claim 1, characterized in that, Step 1) Specifically, information sources are social media platforms, media websites, and local government websites related to earthquakes. Using web crawlers, earthquake disaster keywords are selected for optimization to obtain post-earthquake network data, thereby establishing the original post-earthquake network dataset.

3. The method for dynamic monitoring of post-earthquake disaster emergency response based on network big data as described in claim 1, characterized in that, Step 4) Specifically, the structured dataset is analyzed using TF-IDF text keyword extraction technology.

4. The method for dynamic monitoring of post-earthquake disaster emergency response based on network big data as described in claim 3, characterized in that, In step 4), the TF-IDF value is used to determine whether the indicator is unidentifiable or identifiable.

5. The method for dynamic monitoring of post-earthquake disaster emergency response based on network big data as described in claim 1, characterized in that, Step 6) The information extraction process is as follows: Read a record of an indicator from a dataset of sub-indicators over a certain period of time. For the i-th statement in the read indicator record, use Jieba word segmentation technology to segment the words, convert the segmented words into strings, traverse the strings and compare them with the data dictionary of the corresponding indicator, and output the statements containing the data dictionary to a description field of the result dataset. The statements saved in the description field will then have information extracted according to both numerical and textual descriptive indicator models. The extracted information will be saved to a target field in the result dataset. The numeric type uses the isdigit() function to extract values ​​from the numeric set of the description field in the result dataset; the descriptive type extracts information from the textual descriptive set of the description field in the result dataset, which is achieved by overlaying n statements containing the indicator dictionary; the target data is extracted according to the textual descriptive indicator model, and the extracted data is stored in the target field of the result dataset.

6. The method for dynamic monitoring of post-earthquake disaster emergency response based on network big data as described in claim 1, characterized in that, Spatial visualization and cartographic output includes: Based on the results data table, the fourth-level place names and POI vector points are linked step by step, and the geographic latitude and longitude coordinates of the event are added to establish an earthquake emergency monitoring vector point file; By associating the values ​​of multiple earthquake emergency monitoring indicators with the corresponding vector data, a vector layer with geographic coordinates for the earthquake emergency monitoring indicators is obtained. Quantitative numerical mapping of vector data, i.e., outputting a spatial distribution map of earthquake emergency monitoring.

7. A system based on the post-earthquake disaster emergency dynamic monitoring method based on network big data as described in claim 1, characterized in that the system... include: The system includes modules for establishing a post-earthquake network dataset, data decomposition, quantitative indicators, multi-source databases, predictive model building, spatialization of emergency monitoring data, and visualization and mapping of earthquake emergency monitoring data. The post-earthquake network dataset creation module is based on raw data collection and creates a raw post-earthquake disaster dataset. The data decomposition module is used to integrate data in an attribute structured manner, decompose the collected records in chronological order, extract each column of fields, and establish a standardized database. The quantitative indicator module is used to establish emergency monitoring indicators and obtain multiple sub-indicator datasets including emergency indicator keywords; the multi-source database module is used to establish a cloud-based geographic information database that integrates social, economic and natural multi-source data of disaster areas. The prediction model building module is used to build estimation prediction models for network-unidentifiable indicators, including the prediction model for disaster-affected population indicators. The emergency monitoring data spatialization module is used to perform spatial statistics on earthquake emergency monitoring vector data indexed by time intervals, and to construct earthquake emergency monitoring time series change data. The earthquake emergency monitoring data visualization and mapping module is used to display the spatial distribution characteristics of disasters through mapping and to further detect the environmental conditions of disasters.

Images