Rappresentazione di opere d'arte nel social semantic web: modelli computazionali basati su ontologie per l'integrazione della dimensione emotiva

This thesis project is part of the interdisciplinary research area of Digital Humanities and its main objective will be the study, design and development of ICT solutions for the management of cultural resources, in order to make them accessible and usable in an innovative way by a large audience, with diverse competencies and interests. To achieve this goal, the project intends to investigate the integration of two perspectives: (a) Using semantic technologies (ontologies, Semantic Web resources, Linked Open Data) to represent heterogeneous cultural content (textual documents, photos, movies, paintings, sculptures, etc.). The semantic conceptual model identifies people, places, events and relationships, allowing users to ``discover" new paths of access to cultural heritage. (b) Use of affective computing tools and sentiment analysis. These tools have proven to be helpful in extracting information about opinions and mood from textual data in various application areas, such as social media marketing, political and social analysis. Aspects related to sentiment are also strongly related to the aesthetic experience as recognized by philosophical and psychological theories. Through social media, language feedback on works of art or other cultural resources becomes accessible, and this makes it possible to explore the emotions evoked by resources. The aim of the thesis is to investigate the relationship between sentiment, emotions and semantic description of contents in the field of cultural heritage, with particular emphasis on artistic objects. An analysis framework is presented, that uses the tag as a source of information or other textual footprints that visitors leave to comment an artwork on the social web platforms and returns them associated with the artwork. In order to extract emotional semantics from user tags or comments, the use of methods and tools from different disciplines, from Semantic and Social Web to NLP, is investigated. The idea is that these disciplines provide the basic ingredients for creating a social and semantic space where artworks can be accessed and dynamically organized with reference to an ontology of emotions. NLP methods and resources have been exploited to extract an emotional semantics shared among individuals, which encodes the meaning they perceive and the reactions to art collection they show. This semantics is represented by an ontology of emotions inspired by the dimensional model of human emotions proposed by Plutchik, and subsequently suitably lexicalized with terms of the Italian language. The affective categorization model and the output of emotional analysis are represented by using W3C ontological languages, with the dual benefit of allowing a system of inferences on detected emotions and related artworks, and promoting interoperability and integration of tools developed in the Semantic Web and Linked Data communities. The proposed framework has been implemented and evaluated through its application on a case study in a real domain, i.e. a multimedia art dataset tagged by an online community, part of the ArsMeteo online collection, for which was implemented both an interactive interface, to allow the user to visualize and interact with the result of emotional analysis of artworks, and a SPARQL endpoint to query the ArsMeteo dataset of emotionally enriched artworks and artists. Subsequently, the proposed implementation was evaluated through a user study.