Leveraging Sentiment and Topic Detection Analysis for Decoding Player..
summary
Uses a mixed pipeline: start with manual qualitative coding on a 1,000–2,000 comment sample, then fine‑tune a token‑based NLTK sentiment intensity analyzer and an aspect‑based classifier, validate and expand topics with LDA + OpenAI, and run automated monitoring via a competition analysis scraper that pulls YouTube comments and engagement ratios. Detects polarizing issues (combat difficulty split roughly 50/50), drives a concrete change—eased boss fight mechanics and removed some Souls‑like elements—and reports that the tweak preserves skilled players’ experience.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:50Sentiment and topic analysis
- 3:36Analyzing player feedback
- 5:44Challenges in analysis
- 8:12Scar Above case study
- 14:11Sentiment analysis methods
- 18:45Manual topic coding
- 21:51LDA with OpenAI
- 24:21Combining methods
- 26:49Competition analysis tools
- 27:53Key takeaways
related talks
57mTackling toxicity: How to build a thriving player community
53mBalancing gameplay of aRPG using player behavioral patterns
40mStop Toxicity in Online Games | Gaming Psychology
08mAh-Aloud Method to Comprehend Time-Series Emotion Observation During Gameplay: An Initial Investi...
18mCreating Safe & Thriving Communities: Powerful Content Moderation Strategies......
40mModulating Game Difficulty with Machine Learning
21mMining Controller Inputs to Understand Gameplay
03mExploration of Player Emotions, Behaviours, and Individual Differences across Game Difficulty...