Mining graphs from travel blogs: a review in the context of tour planning

The tremendous expansion of user-generated content has made travel planning a tedious activity. The information overload disorients some travel knowledge seekers in choosing preferred places and visiting them in an appropriate order. Hence, a trip planning system mainly deals with collecting popular points of interests (POIs) in a region and proposing routes accordingly. In contrast to the various data sources of tour planning, travel blogs have been paid less attention. The massive text contained in these narratives manifests bloggers’ interactions with places that have been modeled with different graph representations. This review concentrates on the problem of extracting this graph-structured information from travel blogs with an explicit focus on POI graph, the fundamental trip planning structure that facilitates travel decision making. The study first discusses an analysis of travel blog mining and its dominant research themes, based on which three distinct categories of graphs are proposed. Each category of graphs is further explored using practical cases followed with a comparative analysis of existing works including their techniques, dataset and evaluation metrics. With the underlying idea of enhancing travel planning process with better POI graphs, the existing literature is critically assessed at the end along with a number of promising directions for future research.