Semi-supervised biomedicine event extraction method based on co-training

Abstract

The present invention relates to a semi-supervised biomedicine event extraction method based on co-training. Automatic extraction of biomedicine events greatly interests people with rapid increasing of biomedicine literatures. The scale of the marked biomedicine event corpus is small to influence the performances of the classification algorithm and even cause overfitting. The method provided by the invention identifies more accurate positive instances from unmarked data to enlarge a marked training set. The method comprises the steps of: designing abundant features for usage of an SVM; learning short sentences based on word embedding from Word2vec and Pubmed; further extending the short sentences to dependent short sentences between triggering words and parameters, and inputting the dependent short sentences into a CNN; and finally, performing backfill of samples, meeting conditions, predicted by the SVM and the CNN in the unmarked corpus into the training set, incrementally extendingthe training set. Lots of experiment results show that the new semi-supervised biomedicine event extraction method can effectively extract events.

Description

technical field [0001] The invention relates to the field of text mining, in particular to a method for extracting semi-supervised biomedical events based on co-training. Background technique [0002] Biomedical event extraction is an important branch of information extraction. With the rapid growth of biomedical literature, researchers need a lot of energy and time to acquire relevant scientific knowledge. Therefore, automatic extraction of biomedical event information has attracted great interest. Therefore, it is necessary to extract biomedical events in an efficient and accurate way. [0003] At present, the methods of event extraction can be roughly divided into two categories: rule-based methods and machine learning-based methods. A rule-based event extraction system consists of a series of rules, including sentence structure, grammatical relationship, and semantic relationship. These are defined manually from the training data or learned automatically. Human inte...

AI Technical Summary

Problems solved by technology

[0006] In order to solve the problem that the amount of labeled samples in the existing supervised learning for biomedical event extraction is very small and easy to cause overfitting, the present invention proposes a semi-supervised biomedical event extraction method based on co-training. The content of the invention mainly includes : Use the method of semi-supervised learning to expand the idea of ​​​​annotated sample sets; use the SVM classifier and CNN classifier to train together, select the process of sample backfilling to the training set; the process of constructing short sentence sets used as CNN input; the construction of CNN network Process; sample selection strategy for backfilling into training set in unlabeled sample set

Images