Department of

Computer Science and Engineering

Capstone Projects Details

An Attention- based Medical NER in the Bengali Language

Contributors:
  • Sakila Mahbin Zinat
  • ID: 16172103283
  • Intake/Section: 35/1
  • Tanvir Islam
  • ID: 16172103408
  • Intake/Section: 35/1
  • Zakir Hossain Zamil
  • ID: 16172103305
  • Intake/Section: 35/1
  • Aynur Nahar
  • ID: 16172103300
  • Intake/Section: 35/1
  • Shamima Sukhi
  • ID: 16172103373
  • Intake/Section: 35/1

Abstract:

Medical Named Entity Recognition is a process where medical entities are identified for extracting keywords in particular tasks in the medical sector such as summarizing prescriptions, identifying diseases, etc. NER can make a context more comfortable to understand by identifying entities in the context. In the Bengali language, there is no artificial work that can identify automatically which kind of medical specialist a patient needs to consult based on patients’ problems and symptoms. In this paper, NER has been selected and proposed an attention-based BiLSTM-CRF model for the task of telemedicine consultancy where the patient tells their problems, symptoms, and diseases at the first attempt, both the consultant and patient needs to understand which specialist the patient requires according to the problems or symptoms. This task has been implemented based on a self-made medical dataset in the Bengali language which gives an F1 score of 95.6% accuracy level and performs more efficiently in this task.

Get Full Document