This repository contains "Detecting Dalhousie's Official Twitter Account's Main Interests", as the final project of Advanced Topics in Natural Language Processing in spring 2020. There is a scientific report on this project in this repository. This project fundamentally is a stochastic look to the Twitter accounts of Dalhousie University.
In recent years, Twitter has been a good source for getting updates about the owners of the accounts. Dalhousie university, as a pioneer university in Atlantic Canada, has several twitter accounts that having general knowledge on the topics and main interests of them may be helpful and also entertaining for the ecosystem of the university. This project has a couple of science-backed research on Dalhousie's Twitter accounts, including topic modeling using the Latent Dirichlet Allocation method, polarity calculation, and data redundancy calculation. The insight derived from this research can give a big picture of the atmosphere of social media accounts of the university. According to the results of this research, which is conducted on a dataset gathered before COVID-19 pandemic, events and gatherings are the most popular topics in the tweets of Dal by 19.14 percent. Additionally, it can be calculated that retweets make up 34.9 percent of the tweets of the Twitter community of Dalhousie, which is chosen to be observed in this research.