Automatic construction of POI address lists at city streets from geo-tagged photos and web data: a case study of San Jose City

Abstract

Points of Interest (POIs) are crucial data sources for location based applications. Social media and traditional web can include up-to-date POI information that is emerging in the real world and shared by users and organizations. They have been demonstrated as the potential data sources for enriching the existing POI databases and online map services. Commonly, a POI is associated with a street for navigation, accessing, indexing and searching. The association between a POI and a street is present in the POI address. This paper proposes a novel approach for automatically constructing POI address lists at streets of a city from geo-tagged social media photos and web data. The proposed method can yield POI addresses that are missing on Google Maps, OpenStreetMap or Wikimapia. As a result, it is potentially applied for enriching POI data and enhancing online digital map services. In our approach, we first specify the relation between a POI name discovered from geo-tagged photos and related streets, candidate addresses; and then we utilize this relation to mine the POI address from web snippets by a search engine. We present a case study of San Jose City, California, USA. The analysis results have demonstrated the effectiveness of the proposed method, providing a promising solution for automatically constructing POI address lists at city streets from geo-tagged social media photos and web data.