Bridging the gap between user generated spatial content and the semantic web.

Abstract

This work concerns the development opportunities for Geographical Information coming from new technologies and ongoing web applications. Geographical content created online by users is growing in amount and heterogeneity. OpenStreetMap is the prominent project collecting geodata on the web through users’ contributions. The usability of collected geodata is targeted in the present work. The target users of the data collected online are intended to be the people who plan everyday life decisions spatially, therefore local knowledge on amenities and facilities is considered valuable. Without strict standardization, online communities can create databases affected by low thematic accuracy by using synonymic terms. From another perspective the OSM community standardization efforts may not be known by the generic data consumer wanting to extract information on the location of everyday places. Similar problems have been encountered in the efforts to merge heterogeneous databases. In that environment semantic conflicts arise when heterogeneous databases containing similar objects have to be merged. In particular there is a naming conflict when synonymy and homonymy can affect the merging process. The solution to naming conflict resides in the semantic description of data and in the use of a linguistic resource able to manage synonymy. The perfect environment for the solution of the outlined problem is therefore an environment where languages are able to express semantic relations between data and where a resource exists where synonymy can be handled through semantic relations between nouns. This environment is the Semantic Web. In the semantic web semantic resources from almost every area of human knowledge are published and linked semantically. Therefore in the present work the semantic technologies, tools, resources and ongoing initiatives involving OpenStreetMap have been analyzed and compared. Geographic Information research is already focused on semantic technologies to gain advantage from it in a multiplicity of ways. Therefore some examples of ongoing research on semantic technologies and Geographic Information have been shown. The semantic translation of OpenStreetMap, LinkedGeoData, is thus evaluated and compared to examples of Geographic Information applications involving semantic technologies. To solve the outlined naming conflict semantic technologies and semantic web-based geographic and linguistic resources have to work together and their matching is not an easy task. The linguistic resource used to solve the naming conflict is the semantic translation of WordNet, the prominent linguistic database originally published by Princeton University. The naming conflict has been tackled creating a semantic query expansion that by surfing the web matches a queried term with the linguistic semantic resource and then identifies geo objects originally published online by OpenStreetMap users. Two aspects coexist in the present work. In the short term, the query expansion has been developed and evaluated. In a broader perspective, all the evaluations performed for the integration of geographical user generated content with semantic web technologies and resources will be useful to explore the possibility of further information source integration.