A Biological Test Questions Naming Entity Recognition Method for Fusion Triggers

Training a neural model for named entity recognition (NER) in a new field often requires additional human annotations (for example, a large number of tagged instances), which are often expensive and time-consuming to collect. Therefore, one of the key research problems is how to obtain the supervision effect in an economical and effective way. In this article, we use “entity triggers” for the biological domain, which can help explain how to effectively learn the label of NER's model. Entity triggers are defined as words or words in a sentence, which helps explain why people recognize entities in a sentence. For biological disciplines, according to the characteristics of the high school biology field test data, in a small, small amounts of annotation and trigger a small entity tagging of self-built corpora in the biology one experiment, the experimental results show that using “triggers the matching network can automatically learn” trigger said and soft matching module, which can be easily generalization for the tag will be invisible to the sentence, then use the two-way coding BiLSTM, finally with the airport marking label according to conditions. The experimental results show that this model has a better recognition effect than other algorithms, and can effectively solve the problems of entity recognition difficulty and low accuracy in the task of named entity recognition in the field of biological science, such as insufficient labeling data, and the special ranking of biological science