wuesten/sentiment-analysis-fh-kiel

Sentiment Analysis

For an university project at the University of Applied Sciences Kiel, we conducted a sentiment analysis with the aim of classifying restaurant ratings. The model of "nlptown/bert-base-multilingual-uncased-sentiment" was used as a pre-trained transformer. As a starting value, an accuracy of 63% was already achieved. Based on this, the transformer was fine-tuned to yelp ratings from Hamburg. After that, an accuracy of 82% was achieved.