Semantic profiling and destination recommendation based on crowd-sourced tourist reviews

Abstract

Nowadays tourists rely on technology for inspiration, research, booking, experiencing and sharing. Not only it provides access to endless sources of information, but has become an unbounded source of tourist-related data. In such crowd-sourced data-intensive scenario, we argue that new approaches are required to enrich current and new travelling experiences. This work, which supports the “dreaming stage”, proposes the automatic recommendation of personalised destinations based on textual reviews, i.e., a semantic content-based filter of crowd-sourced information. Our approach relies on Topic Modelling – to extract meaningful information from textual reviews – and Semantic Similarity – to identify relevant recommendations. Our main contribution is the processing of crowd-sourced tourism information employing data mining techniques in order to automatically discover untapped destinations on behalf of tourists.