Research On Knowledge Map Building Technology In Military Equipment Domain Based On Deep Neural Network

In recent years,with the deep learning technology achieving better results in text mining tasks,for the difficult tasks in the equipment knowledge map,we can try to study how to use the basic theory of deep learning to the equipment entities and relationships contained in the text.Aiming at the traditional knowledge-based maps,there are a large number of manual operation preprocessing,feature engineering,heavy manual dependence,and knowledge-based questions and answers rely heavily on the syntactic parsing of questions,etc.This paper is based on deep neural network.The main research contents and innovations of the knowledge map construction technology in the field of military equipment are as follows:A multi-source data acquisition method for equipment field is proposed.Multi-source data collection methods extract,parse,and transform data from multiple sources,and finally represent and store it in a uniform format.This article collects data from military equipment professional websites,search engines,and Wikipedia websites,providing powerful data support for the next analysis and mining of valuable data.A military equipment entity relationship extraction technique based on remote supervision and deep learning is proposed.In this paper,the combination of neural network and attention mechanism is used to extract the entity relationship of the equipment,and remote monitoring is used to obtain the large-scale training data set.The method of remote supervision avoids the time and effort spent manually on building the data set.The GRU model overcomes the problem that the traditional deep learning model can not solve the long-distance dependence,and introduces the attention mechanism at the sentence level,effectively controlling the problem.The impact of redundant data on experimental results.A prototype system of equipment knowledge maps is constructed.The equipment knowledge map prototype system constructed in this paper mainly acts on the algorithm-oriented testers and the users of weapon-oriented knowledge maps,and uses the corresponding data extraction model to establish the map through real data.The main functions of the system include:user identification,data acquisition and processing,and relationship extraction.In summary,this paper focuses on the construction of knowledge maps in the field of military equipment,and studies the data collection of multi-source equipment.Using the basic theory of deep neural networks,the research focuses on the extraction technology of equipment entity relationships.Finally,a more practical knowledge map of equipment field is constructed.Prototype system,these technologies and achievements have certain theoretical significance and application value for the intelligence of equipment information system and the development of sensory equipment.