Detecting Hidden Relations in Geographic Data
Ngoc-Thanh Le, Ryutaro Ichise, Hoai-Bac Le · 2010
Abstract—The amount of linked data is growing rapidly, and so finding suitable entities to link together requires greater effort. For small data sets, it is easy enough to find entities in the data sources and link these together manually; however, doing so for large data sets is impractical. For large sets, a way is needed to discover entities and connect them automatically. In this paper, we present an algorithm to detect hidden owl:sameAs links or hidden relations in data sets. Since geographic names are often highly ambiguous, we used data sets comprising geographic names to implement and evaluate our algorithm. We experimentally compare our algorithm with a naı̈ve algorithm that only uses a URI’s name feature. We found that it is more accurate than the naı̈ve algorithm in most cases, especially for resources in which there is little matching information about features.