An Ontology for Narratives

One of the main problems of the current Digital Libraries (DLs) is the limitation of the informative services offered to the user who aims at discovering the resources of the DL by queries in natural language. Indeed, all DLs provide simple search functionalities that return a ranked list of their resources. No semantic relation among the returned objects is usually reported that could help the user to obtain a more complete knowledge on the subject of the search. The introduction of the Semantic Web, and in particular of the Linked Data, has the potential of improving the search functionalities of DLs. In this context, the long-term aim of this thesis has been to introduce the narrative as new first-class search functionality. As output of a query, the envisaged new search functionality should not only return a list of objects but it should also present one or more narratives, composed of events that are linked to the objects of the existing libraries (e.g. Europeana) and are endowed with a set of semantic relations connecting these events into a meaningful semantic network. As a necessary step towards this direction, the thesis presents an ontology for representing narratives, along with a tool for the construction of narratives based on the ontology. Moreover, it has used to the tool for evaluating the ontology in the context of an experiment centred on the biography of the Italian poet Dante Alighieri, the major Italian poet of the late Middle Ages. More specifically: - An overview of the related works developed in the Semantic Web field and in Narratology, and especially in its sub branch named Computational Narratology was reported. The basic principles of Narratology and Computational Narratology have been reviewed along with the study of the Artificial Intelligence literature, especially of the Event Calculus theory, in order to identify the formal components of narratives. - A conceptualization of narratives has been developed, based on notions derived from narratology and Artificial Intelligence. According to this conceptualization, a narrative consists of a fabula, i.e. the events of a story in chronologically ordered, and several narrations of this fabula (plots), linked to the fabula by an event association relation. A mathematical expression of the conceptualization has been given, in order to provide a characterization of the conceptualization as clear and as precise as possible, also to be used as a basis for the subsequent development of an ontology of narratives, encoded in OWL. The proposed conceptualization has been validated by expressing it into an existing ontology, the CIDOC CRM, and by endowing it with provenance knowledge, also expressed in a derivation of the CRM, named CRMinf. This expression has been used in the validation experiment, consisting in the modelling a narrative of the biography of Dante Alighieri, provided by a biographer who has scientifically supported this research. - The population of the created ontology has been performed by means of a semiautomatic approach implemented by a tool for the construction of narratives which obey the ontology. This tool retrieves and assigns URIs to the instances of the classes of the ontology using Wikidata as external resource and also facilitates the construction and contextualization of events, and their linking to form the fabulae of narratives. - Finally, a qualitative validation of the developed ontology has been carried out. This validation has regarded the evaluation of: (i) the representational adequacy of the ontology by a Dante Alighieri's expert; (ii) the effectiveness of the narrative building tool; (iii) the satisfaction of the users' requirements defined at the beginning of the study. To prove the last point, initial requirements representing pre-requisites of this work have been satisfied by demonstrating that a SPARQL query can be always built to extract the requested information from the knowledge base embodying the narrative.