Attention-Based Multitask Model for Name Entity Recognition and Intent Analysis of Online Medical Questions

Abstract

Abstract: In recent years, various methods of deep learning have been explored and are applied for solving Named Entity Recognition task which are considered as one of the prior task of Natural Language Processing.

Named Entity Recognition has been recognized as a sub task of Information Extraction, where it recognizes and classifies proper noun into their given pre-defined categories like persons, location, organization, date and time etc. The survey mainly makes a focus on different methods of deep learning, which have been approached for NER task to work well along with relative multitask learning techniques to opt a novel model based on Neural Network architecture by performing sequence tagging and text classification for witnessing Named Entity Recognition task and Intent Analysis task for online medical questions.

Both the attention mechanism and multitask learning have been improved the performance of their respective tasks. This method has achieved superior performance in both Name Entity Recognition (NER) and Intent Analysis when compared with other methods. The present method is considered as light-weighted solution that can be suitable on every small server for its deployment. By making use of both the tasks a simple question-answering system has been developed.

Index Terms: Name Entity Recognition, Intent Analysis, Attention mechanism, Multitask neural network, Sequence tagging, Text Classification.