sentiment analysis tool for the video game inustry (satvgi) · group 24 - kyle thakker, sean hwang,...
TRANSCRIPT
Group 24 - Kyle Thakker, Sean Hwang, Aaron Hu, Sagar Phanda{klt117, sh1060, ajh193, sp1412}@scarletmail.rutgers.edu
Advisor: Prof. Jorge Ortiz
Goal Accurately determine consumer sentiment towards a subset of
video games by scraping user comments from a social media site
Design and develop a web application allowing users to view sentiment analysis metrics for these games
System
References[1] https://spring.io/guides[2] https://praw.readthedocs.io/en/latest/[3] https://react-bootstrap.github.io/[4] https://reactjs.org/docs/getting-started.html
Acknowledgement
We would like to thank our advisor, Prof. Jorge Ortiz, for his help and guidance
Sentiment Analysis Tool for the Video Game Inustry (SATVGI)
Research Challenges Finding a useful and sizable data set to train our classifier
Accurately judging which posts have comments that are relevant enough to be included in our sentiment analysis
Database
Stores sentiment and game data
ScraperScrapes and classifies Reddit comments as
either positive ornegative using the Naive
Bayes Algorithm and stores results in a database
Web Application
FrontendDisplays Data
to the user
BackendREST API handling
transfer of databetween frontend
and database
Scraper
Sentiment Data Source
Results
[1] Home Page
[2] Game Data Page
Motivation
Developers and publishers need to understand the public’s feelings towards their games, especially after games are updated or news is released
Consumers are able to make more informed purchasing decisions when they understand the feelings of the public toward the product they are considering buying
Existing sites that aggregate reviews based on numeric scores can be problematic as reviewers, especially users, do not always utilize the same numeric scale
Consumers are often very vocal about their feelings towards games on social media sites
We found that users spoke more positively about games belonging to long existing franchises with loyal fan bases
We found that for all games, the majority of comments were negative, though the degrees of negativity vary