Linking Spatial Named Entities to the Web of Data for Geographical Analysis of Historical Texts

Abstract

In our work, we are interested in facilitating the exploration by scholars of the geography of texts: in particular, historical narrative texts describing routes. Semantic annotation constitutes the first step to enrich such text with the necessary information for producing analytical maps. The present article focuses on the disambiguation of spatial named entities (SNE) by the attribution of an identifier of the ever-growing Web of Data. This giant knowledge base (KB) provides qualitative spatial information about geographic entities, in particular spatial relations such as (:Paris :southOf :Lille), (:Paris :country :France). We thus propose a graph-matching algorithm relying on the A* algorithm and graph-edit distances for choosing the best referent in the KB for the SNE. We performed preliminary experiments and noted the clear gain in performance. We propose some examples of maps that are built semi-automatically. Finally, we draw conclusions and describe our plans of future work.