Sentiment analysis of comments received online

Inspiration

In a flood of bot accounts and fake reviews I thought if my opinion really mattered when I was asked to give a feedback on an app, then I thought I should make something for everyone's opinions to matter and make it simpler to understand the overall opinion of the people.

What it does

when many feedbacks are sent through online forums most of them include spam or repeated content ,so I wanted to make a classification model that clearly distinguishes positive,negative or ambiguous reviews.

How we built it

Challenges we ran into

when a para of information was given , the model considered it as one big statement and printed whatever repeated more times(i.e if a para had 3 positive and 2 negative statements then the output was positive). I dealed with this problem by adding a mixed statement option.

Accomplishments that we're proud of

What we learned

What's next for Sentiment analysis of comments received online

due to less time I had to rush the model but I want to improve it by adding seperate web scraping, data cleaning and self trained model to the project. The ui too is a bit plain and I have to add more data visuals like word cloud for simpler analysis of comments.

Built With

• css
• fastapi
• html
• javascript
• matplotlib
• python
• sentencepiece
• spacy
• torch
• transformers
• uvicorn
• wordcloud