Full text
| |
| Author(s): |
Hugo Augusto Alves
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | Campinas, SP. |
| Institution: | Universidade Estadual de Campinas (UNICAMP). Instituto de Computação |
| Defense date: | 2012-04-16 |
| Examining board members: |
André Santanchè;
Maria da Graça Campos Pimentel;
Ricardo da Silva Torres
|
| Advisor: | André Santanchè |
| Abstract | |
An increasing number of web repositories relies on tag-based metadata to organize and classify their content. The users of these systems freely associate tags with resources of the system - e.g., URLs, images, and bookmarks. The term folksonomy refers to this collective classification, which emerges from tagging carried by users interacting in web social environments. One of the major strengths of folksonomies is their simplicity due to the absence of a controlled vocabulary. Folksonomies grow organically, reflecting the knowledge of a community of users. On the other hand, this lack of structure leads to difficulties in operations of content organization and discovery. Better results can be obtained if we take into account the semantic relations among tags. For this reason, many proposals were developed aiming to relate ontologies and folksonomies, combining the systematized structure of ontologies to the latent semantics of folksonomies. While in one direction some approaches build "social ontologies" from folksonomic data, in the other direction some approaches connect tags to existing ontologies. In both cases they are unidirectional approaches, i.e., one model is used only to support the enrichment of the other. Our proposal, on the other hand, is bidirectional. Ontologies and folksonomies are fused in a new entity, we call "folksonomized ontology", which combines complementary aspects of both. The formal and engineered knowledge of ontologies is fused with the latent semantics of social data. In this dissertation we present our folksonomized ontology and its outcomes. We introduce here a formal framework to analyze the related work, confronting it with our approach. Besides the improvements in indexing and discovery operations, which are validated by practical experiments, we propose a 3E Steps technique to support ontology evolvement by using folksonomic data. We also have implemented a tool prototype to build folksonomized ontologies and to support ontology review (AU) | |
| FAPESP's process: | 10/14217-3 - Integrating and enriching folksonomies and ontologies |
| Grantee: | Hugo Augusto Alves |
| Support Opportunities: | Scholarships in Brazil - Master |