Using Bridging Knowledge to Expose Meaning and Structure in Folksonomic Tags

The burden of manually developing and maintaining ontologies to support the changing requirements of real-world applications represents a substantial barrier to the successful deployment of advanced information exploitation and knowledge management technologies. One promising approach to overcoming these challenges is to exploit the knowledge that is arising within Web 2.0 applications. These tools gather huge volumes of human insight by allowing users to organize multimedia information in a nearly effortless manner according to their own subjective preferences. However, researchers have only demonstrated a limited ability to automatically extract semantic knowledge from these resources. We propose to significantly improve the accuracy of tag enrichment processes through the automated extraction of relational information from news and Web text. The relational triples (x y) produced by this Open Information Extraction process will act as a bridge between incomplete and static ontologies and unstructured, dynamic folksonomies. This integrated approach will allow us to achieve much higher levels of recall than state of the art approaches, and will ultimately allow for the largely automated adaptation of ontologies. Our Phase I prototype will provide a solid foundation for the complete implementation of this semantic enrichment tool in Phase II and its eventual commercialization.