Self-Annotation for fine-grained geospatial relation extraction

A great deal of information on the Web is represented in both textual and structured form. The structured form is machinereadable and can be used to augment the textual data. We call this augmentation – the annotation of texts with relations that are included in the structured data – self-annotation. In this paper, we introduce self-annotation as a new supervised learning approach for developing and implementing a system that extracts finegrained relations between entities. The main benefit of self-annotation is that it does not require manual labeling. The input of the learned model is a representation of the free text, its output structured relations. Thus, the model, once learned, can be applied to any arbitrary free text. We describe the challenges for the self-annotation process and give results for a sample relation extraction system. To deal with the challenge of finegrained relations, we implement and evaluate both shallow and deep linguistic analysis, focusing on German.