Proposal of time-crawler which collects an event time by reading exif data in blogs

To enable an outdoor user to acquire mobile specific information, an information retrieval system should provide contents that reflect user's circumstance such as user's position and time. There are a lot of information services that can provide contents based on geographical location. But, about temporal information, there is few works, because there is no useful indexing method for time information. In this paper, we propose a crawler which collects temporal expressions by reading Exif data of photos on blogs. Because the Exif data has shooting time of photo, the proposed system is expected to have good precision with small granularity. As a result of evaluation, the precision of the proposed method reached 0.72 even though the proposed method adopts a simple algorithm.

[1]  Inderjeet Mani,et al.  Robust Temporal Processing of News , 2000, ACL.

[2]  Leonard Ray Teel The Weather Channel , 1982 .

[3]  Takashi Inui,et al.  Time Period Identification of Events in Text , 2006, ACL.

[4]  Shinji Shimojo,et al.  MapWiki: A Map-based Content Sharing System for Distributed Location-dependent Information , 2006, J. Comput..

[5]  Donna Gates,et al.  Understanding Temporal Expressions in Emails , 2006, NAACL.

[6]  Xiaofeng Meng,et al.  Postal Address Detection fromWeb Documents , 2005, International Workshop on Challenges in Web Information Retrieval and Integration.