Project information

  • Category: Web
  • Technologies: React, FastAPI, sklearn, spacy, Heroku
  • Project date: Summer 2020
  • Project URL: drive.google.com

Targeted Sentiment Analysis

One application that we are trying to tackle in this project is analyzing the sentiment of a corpus in relation to the subject of the text. For example, given a sentence that references multiple people, what is the sentiment of the sentence towards only one of the people mentioned? This is different from the typical sentiment analysis which does not consider only the text that is directed at the subject, but rather every word in the sentence. We want to leverage dependency parsing and other methods of determining the subject of a sentence to calculate the sentiment of this corpus towards the subject. We will combine this with word embedding and machine learning to hopefully achieve a system that can both recognize the subject and calculate the sentiment towards that subject.