Biomedical semantic relation extraction method based on multilayer neural network and external knowledge base

Abstract

The invention provides a biomedical semantic relation extraction method based on a multilayer neural network and an external knowledge base, aims to extract an entity-relation pair from a biomedical text, and provide technical support for mining mass biomedical text data and constructing a biomedical relation network. The semantic relation extraction method based on the multilayer neural network is adopted, the model can repeatedly extract effective information in the text through the multilayer neural network structure, the problem that the information extraction capacity of a traditional neural network is limited is solved, and the classification performance of the model is improved. A good effect is achieved on different data sets, and the semantic relation can be efficiently and accurately extracted from a large number of biomedical texts. Knowledge in an external biomedical database is reasonably introduced; a UniProtKB database aiming at a single entity and a BAR and IntAct database aiming at a binary relation are matched with an Attention mechanism to effectively screen single entity information, so that the utilization effect of external knowledge is improved.

Description

technical field [0001] The invention belongs to the natural language processing technology in the field of biomedicine, relates to biomedical text mining, and specifically relates to a biomedical semantic relationship extraction method based on a multilayer neural network and an external knowledge base. Background technique [0002] Semantic relationship extraction is a key step in biomedical text mining. It uses natural language processing technology to automatically extract the relationship between biological entities in massive, unstructured and rapidly growing biomedical literature, which in turn helps to build biomedical semantic relational network. [0003] In relation extraction tasks in the field of natural language processing, neural network models have become mainstream, especially LSTM networks. In the LSTM network, the time series data can be memorized through the input gate, forget gate and output gate while avoiding the loss of key information caused by too lo...

AI Technical Summary

Problems solved by technology

[0005] There have been some works using knowledge base information in relational extraction tasks in the field of natural language processing, usually focusing on a single type of knowledge, such as the description of entities; and the introduction of external knowledge will improve the model due to the lack of context information limited

Images