By extracting keywords and sentiment analysis customers can get a better insight of all the reviews and filter their search on your platform. A tool to extract the most common positive and negative aspects of hotels based on user reviews. It also recommends a hotel to you based on what you have responded positively to in the past. Instead of using history of hotel bookings and client bookings, this API works out of the box for new customers as well.
Consists of 2 parts:
- Keyword extraction: For this there is a supervised and an unsupervised technique, relying on TF-IDF + frequency and TF-IDF + random forest respectively.
- Recommender system: For this there are two techniques, one based on cosine similarity of RAKE, and the other on cosine similarity of TF-IDF.
Contributors:
- Elien van Riet
- David Fortini
- Lu Zheng
- Xu Zhang
- Rudy Schueder