PsycheTagger -- using HMM to predict sentiments over

AK

Project Description

PsycheTagger -- using HMM to predict sentiments over word level

The human elements of personality working behind the creation of a write-up play an important part in determining the final dominant mood of a text. This project builds a tool, PsycheTagger, which extracts the emotive content of a text in English Language in its context and tags each open-class word of the text with one of the predefined psyche categories that represent the emotive content. Working in the lines of statistical Parts-of-Speech Taggers, this tool is an example of a semantic tagger. The tagger self-ranks its choices with a probabilistic score, calculated using Viterbi algorithm run on a Hidden Markov Model of the psyche categories. The results of the tagging exercise are critically evaluated on the Likert scale. These results strongly justify the validity and determine high accuracy of tagging using the probabilistic parser.


Highlights

Received a promotion
Received recognition / award

Difficulties

Management had to be convinced

Products Used

Technical Skills Used

  • C++
  • C#
  • English Language

Technical Certifications

  • Computational Linguistics

Awards

  • WSEAS Journal Publication
  • Lahore31.549774.3436