ReAdvisor: Neural Response Generation and Sentiment Analysis in the Context of Restaurant and Hotel Reviews

Online reviews are extremely important for the tourism industry. Reviews written by customers have a large influence on potential customers' decisions. In addition, these reviews provide important information for a business. In the tourism industry, especially in hospitality (e.g. restaurants and hotels), most businesses do not have the resources required to respond to the many reviews posted on various online platforms.

The project ReAdvisor had the following three objectives:

• provide automatic sentiment analysis of customer reviews
• develop semi-automatic response generation of customer reviews
• optimise the interaction between industry clients and review authors as well as between review authors and the automatic response generation system

The first two of these aims was achieved through a collaboration between the Department of Computational Linguistics and our industry partner (see below).

Zoom

TripAdvisor review

Funding

This project received Innosuisse funding from January 2020 until July 2021 (38943.1 IP-ICT).

Project Partners

 re:spondelligent (main industry partner, project management; Alexander Zaugg, Natalia Zaugg)

welante (industry partner; Markus Welter)

Information Management Research Group (IMRG), Department of Informatics, UZH (research partner; Prof. Dr. Gerhard Schwabe, Dr. Mateusz Dolata, Dzmitry Katsiuba)

Project duration

January 2020 until July 2021

Project head

 Prof. Dr. Martin Volk

Researchers

Dr. Sarah Ebling

Dr. des. Michael Amsler

Tannon Kew