Deploying the semantic Web
Strategic decision making, especially in the growing fields of business and competitive intelligence, requires the acquisition of relevant pieces of information such as market trends, fusions, and company values. It is characteristic that the extraction process of this information from multiple sources is resource intensive and importantly, has to be performed regularly and quite frequently. Within the PARMENIDES project, funded by the Fifth Framework Programme, an ontology-driven systematic approach for integrating the entire process of information gathering and analysis was adopted. For decomposing business into conventional queries, dedicated methods were developed that semi-automatically establish ontologies over specific application domains, annotate documents with ontology components and identify entities within them. Two complementary methodologies were exploited for the establishment of ontologies over unstructured texts, essentially press releases and reports that are not part of a database. The Template-oriented methodology, with its origins in natural language processing, expresses the structure and semantics of documents by means of rule templates and enriches them with formal annotations. On the other hand, the KDD (Knowledge Discovery in Databases) methodology supports the discovery of concepts and relationships characterising the documents. Furthermore, it is relevant to the domain of discourse. Extracted information assets are preserved in a document warehouse to support knowledge discovery tasks over time and to enable the spotting of new trends that would remain hidden otherwise. Although the plethora of "ready to use" ontologies offers standardised conceptual material, often combined with valued suggestions about the associated conceptual relations, this knowledge must frequently be restructured. The Parmenides knowledge extraction system can cover the paucity of ontological material for domains which are currently the focus of numerous research activities.