/AFCC

Automated Feedback Constructiveness Classifier for Product Reviews

Primary LanguageJupyter Notebook

AFCC - Automated Feedback Constructiveness Classifier

Author: Ugo Loobuyck

Course: Master's Thesis in Computational Linguistics

Beginning date: 21st January 2020

End date: May 27th 2020

Project aims:

  • Investigate constructiveness in online reviews:

    • Provide a valid and reliable constructiveness ranking scale.
    • Provide training and test annotated data (~4k reviews).
  • Show the possibility to classify feedback reviews based on constructiveness:

    • Compare different models, different sets of features and number of features.
    • Compare our results with binary classification and out-of-domain data.

Table of contents:

Results and citation when the project is over.