Get into the spirit of a location by mining user-generated travelogues

User-generated travelogues have contributed abundant location representative information (e.g. attracting scenic spots, activities, and local customs), which can greatly facilitate people in trip planning and destination understanding applications. Existing work on location information extraction from user-generated travelogues has primarily focused on the contents, while in this work, we resolve both the contents and structures of travelogues, as well as investigating the interplay of the two. We propose a two-part framework to mine location representative knowledge from travelogues. The first part resolves travelogues in a geographic view. It discovers real location entities subordinated to travel destinations, then decomposes travelogues to acquire corresponding descriptions for each of these entities. Built upon the results of the first part, the second part performs content resolution in a semantic view. By extracting location characterizing concepts and the relatedness among them, it can form a representative concept network for each location entity. Based on a large collection of travelogues, the proposed framework is evaluated using both objective and subjective evaluation methods and shows promising results.