Process - parksjin01/2018-Graduation-Project GitHub Wiki

  1. Gaze tracking
  • Get gazing data (i.e. fixation, saccade, regression)
  • Sending data to server by AJAX
  1. Pattern analyze
  • We defind difficulty as two types
  • First case, sentence include difficult words
  • Second case, grammar or sequence of sentence is difficult

2.1. Sentence include difficult words

  • We identified first case by using fixation.
  • Average fixation is 225ms ~ 250ms.
  • If fixation on specific word is longer than we decide that sentence is in first case.

2.2. Grammar or sequence of sentence is difficult

  • We identified second case by using average fixation, saccade and regression.
  • We set each values threshold heuristically.
  1. Natural Language Processing
  • We find the meaning of difficult words by searching Wiki
  • We analyze difficult sentence by using NLTK and Stanford Parser.
  • We classify which grammar is used in sentence.
  • We split sentence into phrase or clause.
  • Respond result to web browser as AJAX response