Research On The Construction And Intelligent Retrieval Method Of Power Equipment Knowledge Gragh

With the development of the economy and the increase of population,human demand for energy is constantly increasing.As an important component of energy,power equipment bears the important responsibility of meeting the energy demand.The field of power equipment involves a large amount of knowledge and information,including equipment classification,performance parameters,working principles,maintenance,and more.By establishing a knowledge graph,these scattered knowledge can be organized and managed,forming a structured knowledge system that facilitates users in accessing and utilizing relevant knowledge.This paper focuses on researching the challenging problem of constructing a knowledge graph for power equipment and applying intelligent retrieval methods based on the power equipment knowledge graph.The main research contents are as follows:(1)Research on the recognition model of named entities of power equipment.In response to the problem of many long,rare,and domain-specific terms in the entity names of power equipment,a Chinese named entity recognition model based on the fusion of multiple features of characters and words is proposed.By incorporating features of words and word categories in addition to traditional character vectors,the model’s performance is improved.Firstly,the dictionary of relevant terms in the field of electric power is sorted out from the texts related to power equipment on the network,and the text word segmentation and part-of-speech labeling are carried out under its guidance.Then,use Word2 Vec to obtain the corresponding dictionaries of character-character vector and word-word vector,and the word category vector is randomly initialized;Finally,the three feature vectors are integrated in series as input vectors,which are input into the Bi LSTM-CRF model for sequence annotation to obtain the power equipment entity.(2)Research on the extraction model of the entity relationship of power equipment.To address the limitation of traditional GCN models in effectively utilizing different dependency types in the syntactic dependency tree for relation extraction tasks,a power equipment entity relation extraction model based on attention graph convolutional neural networks is proposed.The model vectorizes the syntactic dependency type information into attention matrices and integrates them into the standard GCN.We introduce weights in the GCN to enhance the feature extraction performance of the model.Firstly,process the dataset corpus into a syntactic dependency tree and construct it into a graph structure to form a dependency matrix and a dependent type matrix.Then,different weights are assigned to different dependencies between any two words,and the calculation of weights is based on the connections and their dependency types,forming an attention matrix;Finally,the attention matrix is integrated with GCN to predict the relationship according to the weight of learning.(3)Research on intelligent retrieval methods based on knowledge atlas of power equipment.To address the limitation of traditional information retrieval methods that rely solely on literal keyword matching,a knowledge retrieval method based on the power equipment knowledge graph is proposed for natural language queries.Firstly,using the top-down approach to building power equipment ontology,a knowledge map of power equipment is constructed and stored for visualization based on the results of entity recognition and relationship extraction described above.Afterwards,Chinese word segmentation,problem classification,and question entity recognition are performed on the specific questions in intelligent retrieval to distinguish entities or relationships.Finally,according to the obtained intention,the query sentence is constructed and the specific knowledge base is retrieved to obtain the query results.In summary,this project focuses on the relevant technologies and intelligent retrieval methods for constructing knowledge graphs of power equipment.Firstly,entity recognition is performed on power equipment related texts,and then relationships between power equipment entities are extracted to construct a knowledge graph of power equipment.Finally,intelligent retrieval methods are studied based on the constructed knowledge graph of power equipment,reducing the difficulty of obtaining power equipment knowledge.