Eye Tracking for Natural Language Processing

Published in Indian Institute of Technology, Bombay, 2014

Guide: Prof. Pushpak Bhattacharyya, Indian Institute of Technology, Bombay

  • Analyzed people behavior of carrying NLP tasks and increasing efficiency of NLP algorithms using Eye Tracking
  • Given a Sentence S and N participants for conducting NLP Experiment, our objective is to find the Consensus Eye Movements, using Eye Movements of N participants recorded through eye tracker.
  • Finding the Areas of Interest (AOIs) that are common to all the participants
  • Clustering of Fixations to Determine Areas of Interests
  • Finding an appropriate sequence of the Areas of Interest to synthesize the consensus scanpath
  • Getting AOI’s and combining them to generate Consensus sequence is done by K-mean clustering, Bayesian Network and Viterbi Algorithm