quora-sentiment-analysis

Quora-Sentiment-Analysis

Based on AFINN-111 to do sentiment-analysis on Quora answers and blogposts .

Here’s what i did

• Scraped answers and blogposts from Quora and conveted them into .txt format .
• Ignored
  1. Stopwords
  2. Non-Alpha characters [ # , % , * , ! .. ]
  3. Words which have no sentiment score defined [words , text , answer … ]
• Used the sentimental score-carrying words in AFINN-111.txt to deduce the overall sentiment of an answer or a blogpost .

Samples

• What are some good habits to follow?- Nelson Wang

• Build a solid career in tech without a CS major - Aman Goels

Usage

        python3 main.py  [parameter]

        Parameters:
                pick_answer  - pick answer directly
                pick_profile - pick answers from profile

• Install requirements

pip install -r requirements.txt

Known Issues

• [ ] No priority assigned to words occuring multiple times in the text . ( didn’t get weights to assign :/ )
• [ ] Text format contains words[upvotes , views , Sitemap] which should be part of an answer

• [ ] Ignored words for which scores aren’t defined under AFINN-111.txt .

• Open for PR’s and Issues .

Enhancements

• Some of the words which aren’t included under AFINN-111.txt are ignored ( assigned a score of 0 ) in this case, however we can deduce score for those words . for example , if the word hit always appears in the proximity of words like sad or unhappy , then we can conclude that the word hit carries a negative sentiment . supporting paper

Note

• The answers referenced under samples are picked up randomly .