Creating and exploiting ontologies for semantic annotation of text

An ontology is a description of conceptual knowledge organized in a com-puter based representation. Semantic annotation of text is the process of making semantic information (e.g. about entities and relations) formally explicit. Semantic annotation and ontologies are involved in two main and related tasks: an ontology can be used for semantic annotation, this latter requiring a formal representation of the domain of interest during the process of natural language understanding. On the contrary, texts can be useful sources of knowledge for populating and enhancing ontologies. These two tasks can be combined in a cyclic process: ontologies are used as the "world knowledge" for interpreting the text and assigning the correct meanings to linguistic structures, while semantic annotation provides new knowledge to be exploited for the enrichment and refinement of the ontology itself. This (apparent) vicious circle (between the need of having the domain represented in the ontology for the semantic annotation process and the enrichment of the ontology based on the results obtained from the annotation) can be turned to a virtuous circle if the necessary conditions are set to let the evolving ontology and the annotation process interact in a synergetic way. The construction and maintenance of an ontology can be a very costly engineering process: in order to alleviate the costs several proposals for automatically learning ontologies from data have emerged. In particular "ontology learning from text" has recently become quite popular since textual resources are still the main means for knowledge encoding and transfer used by people. Semantic annotation of text requires the application of natural language processing techniques to reconstruct the syntactic structure of sentences, going through tokenization, morphological analysis and part-of-speech tagging. Once syntax has been made explicit, the ontology can be used to drive the attribution of meaning to the syntactical analysed text. In this work we present the state of the art of ontology learning from text and semantic annotation and the techniques necessary to make optimum use of their symbiosis.