trend

Social Media Topic Sentiment Analysis and Visualisation

A self-hosted web application that allows for user inputted topic sentiment analysis and visualisation of Twitter tweets.

This project was built as the Final Year Project when pursuing the Bachelor of Computer Science (Honours) in Data Science in Tunku Abdul Rahman University of Management and Technology.

Accuracy of sentiments are achieved through the use of RoBERTa-base model by cardiffnlp.

Front-End

The front end is built on the Svelte Framework. tinro was used to handle routing between pages. Various Svelte libraries were used to enhance user experience such as svelte-loading-spinners. For certain animations, animejs was used to animate them. ApexChart.js was used for data visualisation through charts.

Back-End

The back end is built on Python with the Flask library as end-points for interaction between the web application and the back-end. Data pre-processing of tweets utilised Regex expressions, cleantext, and Pandas dataframes.

Tweet scrapping

The twint library is used for scrapping tweets from Twitter.

More Pictures

Limitations

• Lack of queue system prevents multiple concurrent users from utilising the system at the same time. Redis has been considered to alleviate this limitation.