Font Size: a A A

Construction And Application Of Knowledge Graph In The Field Of Debris Flow Disaster

Posted on:2022-02-18Degree:MasterType:Thesis
Country:ChinaCandidate:H ZhangFull Text:PDF
GTID:2480306488960199Subject:Master of Engineering
Abstract/Summary:PDF Full Text Request
Debris flow disasters occur almost every year,bringing threats to people and loss of property.People have made many studies on debris flow disasters,such as the formation,prevention,early warning,and risk assessment of debris flow.With the development of network technology,debris flow-related information is scattered in various places on the network,which will result in people not being able to get accurate and detailed answers when they want to query this information,which is not conducive to people's research on debris flow disasters,and this is not convenient for sharing relevant information in the field of debris flow.This paper constructs a knowledge map in the field of debris flow disasters in response to these problems,and on this basis,this paper designs and implements a knowledge retrieval platform in the field of debris flow disasters.This can promote the sharing of relevant information in the field of debris flow disasters and assist in the prevention and control of debris flow disasters.The main work of this paper is as follows:(1)A Chinese word segmentation model based on HMM+Bi-LSTM is proposed.In order to extract more accurate and better quality information in the field of debris flow disasters,it is necessary to segment the information in the field of debris flow disasters.According to the characteristics of information vocabulary in this field,this paper combines Hidden Markov Model(HMM)word segmentation and Bidirectional Long Short-Term Memory neural network(Bi-LSTM)model word segmentation to improve the word segmentation effect,and the accuracy of the model is verified with the collated debris flow disaster domain corpus.The final result shows that the word segmentation accuracy of the model is compared with HMM Chinese word segmentation model and BI-LSTM Chinese word segmentation model improved by25.17% and 0.12% respectively.(2)An entity extraction method based on HMM+CRF part-of-speech tagging is proposed.In the process of knowledge extraction in the field of debris flow disasters,this paper improves the entity extraction method in the field of debris flow disasters.First,the model add a layer of HMM model to the CRF part-of-speech tagging model,and this paper divide the sentence to be segmented into simple and complex parts.HMM layer and CRF layer respectively mark these two parts of speech,and then The model performs regular expression extraction on the part-of-speech tagging results to obtain entities.Finally,the model uses the collated corpus of the debris flow disaster field to verify the accuracy.The verification result shows the model compared with the separate HMM entity extraction model and the CRF entity extraction model,the entity extraction accuracy rate of the system are improved by 2.2% and 0.3%respectively.(3)A knowledge graph in the field of debris flow disasters is constructed.First of all,this article obtains a large amount of information and data about the field of debris flow disasters from the Internet,libraries and other channels,and organizes them into a corpus of the field of debris flow disasters.Then this paper uses the proposed Chinese word segmentation model and entity extraction model to carry out Chinese word segmentation and entity extraction for information in the field of debris flow disasters.Subsequently,this paper adopts the method based on dependency syntax analysis to extract relations.Finally,this paper defines the entity and relationship types in the field of debris flow disasters,and it saves the extracted entities and relationships in the Neo4 j graph database in the form of triples(entity-relation-entity).(4)A retrieval model based on domain knowledge graph of debris flow disaster is designed.The model mainly includes three parts.The first part is the preprocessing of the query statement,which uses LTP components for word segmentation,semantic analysis,entity extraction and relationship extraction.The second part is the query expansion,which uses Word Net based synonym query and semantic similarity calculation based on homonym forest to analyze the entity or relationship words in the query statement.The third part is to query in the knowledge map of debris flow disaster domain.After users input query statements,the model uses cypher language to query entities or relationships in the knowledge graph.(5)A retrieval system based on the knowledge map in the field of debris flow disasters is designed and implemented.Supported by the well-built knowledge map in the field of debris flow disasters,a search system in the field of debris flow disasters is designed,and the system is designed as a natural language processing module,entity name query module,and natural language retrieval module.Finally,the Django framework and Python is used.Users can experience the three functional modules of the system in the search box through a web browser,and retrieve knowledge in the field of debris flow disasters.
Keywords/Search Tags:Debris flow disaster, knowledge Graph, Chinese word segmentation, entity extraction, knowledge retrieval
PDF Full Text Request
Related items