Video Game Sentiment Analysis

• 

  Cyberpunk 2077, most used review words ("positive")

• 

  Cyberpunk 2077, most used review words ("negative")

Inspiration

With our combined interests in gaming and machine-learning methods, we wanted to better understand video game user reviews before we try out games we have not played yet.

What it does

• Our sentiment analyzer interface does web scraping to take in user review data for a given game through Metacritic.com. It then creates a .csv file for reading the data.
• Dataset is separated by "positive" and "negative" review categories, then the interface generates 2 corresponding word clouds that showcase the most commonly used words in the reviews.

How we built it

• We used Python libraries such as Pandas, langdetect, Flask, and wordcloud. We also used HTML code for our interface.

Challenges we ran into

• Initially, we tried out different programming libraries for sentiment analysis, including HuggingFace and NLTK (Natural Language Toolkit).
• Struggled to debug and get the program running on a Windows terminal, so we had to switch to Linux / Ubuntu for the analyzer to function. There were plenty of errors each one of us faced that we had to overcome.

Accomplishments that we're proud of

• Correctly using different Python libraries and HTML to craft a functioning project!
• Creating a working prototype for sentiment analysis!

What we learned

• Learning how to use new Python programming libraries and how they can be combined for programming useability.

What's next for GamingReviewAnalysis

• Improve the interface so that it is more robust, user-friendly, and useable on any video game review page.
• Use other ML / Deep Learning method(s) to further analyze feelings towards different video games
• Beautify our HTML interface (aesthetics)

Built With

• flask
• html
• jupyter
• machine-learning
• pandas
• python