Research Of The Medical Knowledge Based On Knowledge Graph

The explosive growth of Internet information not only brings a wealth of information knowledge to users,but also makes it difficult for users to screen the required knowledge.Traditional search engines on the Internet full-text index data based on keyword matching index through the way for users to return links to related information,rather than explicit knowledge,users still need to search from a large number of redundant links returned and refine their own knowledge required.How to provide accurate information for users from massive and structured information has become a hot research topic in current knowledge search.The emergence of knowledge graph technology provides a new solution for the study.Knowledge maps can express the information of entities in the real world and the association between entities and concepts in a more intuitive way.This paper takes the Internet text data as the corpus resource,and studies the named entity recognition.This paper studies the problem of sequence annotation in the construction of knowledge map.In this paper,we use the long and short term memory(LSTM)network structure,in order to retain more feature information in training process,we propose a D-LSTM model using pre training word vector and fine-tuning word vector to extend LSTM structure unit.In addition,in order to deal with the sparse text in the medical field,this paper proposes an improved model CTD-BLSTM based on D-LSTM and Co-training semi supervised method,and further improves the recognition efficiency by iterative training.In this paper,CTD-BLSTM algorithm is written in Python,and the contrast experiment between this model and original BLSTM is set up,and the contrast experiment with complete data set training is carried out,which proves that this method has better recognition effect and adaptability.Finally,the knowledge atlas of Chinese medical field is built on the basis of this.Based on the knowledge atlas of the Chinese medical field,the medical knowledge search system is designed and implemented in Java language.The system can identify user's search intention by analyzing user input natural language,syntactic analysis and semantic dependency analysis.With knowledge map,it can return user's knowledge in a more intuitive and precise way.