Assisted adaptation of semantic annotations on RDF bases

Author(s)

Ênio de Jesus Pontes Monteiro

• Document type: Master's Dissertation
• Press: Campinas, SP.
• Institution: Universidade Estadual de Campinas (UNICAMP). Instituto de Computação
• Defense date: 2021-12-13

Examining board members

Julio Cesar dos Reis; Rita Cristina Galarraga Berardi; Leonardo Montecchi

• Advisor: Julio Cesar dos Reis

Abstract

Access to and use of semantically defined meta-data can benefit several types of computational tasks. Data integration between different systems and search engines can explore semantic definitions associated with documents and other web resources to propose ways to combine and find elements of information before not explicitly related to each other. Generating annotations from the growing number of interconnected RDF repositories makes available meta-data needed to enable semantically enriched applications on the Web. For example, semi-automatic tools perform the recognition of entities mentioned as persons and places in Web documents. However, semantic definition of these entities might undergo modifications in their source bases, which implies in a challenging scenario. In this context, generated annotations need to be updated according to the evolution of their underlying knowledge bases used in the definition of annotations. Existing literature has presented methodologies to address this problem, mainly exploring the maintenance of annotations regarding changes at the ontology level (TBox). Our literature analysis did not detect investigations that addressed the maintenance of annotations considering their generation at the level of instance of concepts (ABox). This MS.c. dissertation proposes techniques to recognize the annotations affected by the evolution of RDF datasets. We formalize rules that decide the type of action to be applied to the annotation. We develope a framework capable of identifying and applying adaptation actions to annotations affected by the evolution of RDF knowledge bases. The adaptation comprises the execution of three steps: (1) detection of a series of modifications that occurred in a given time period based on two releases of a RDF dataset; (2) recognition and filtering of annotations affected by RDF changes; (3) application of corrective actions to affected and outdated annotations in an assisted way. Experimental evaluations demonstrate the applicability of the proposal in real RDF bases (AU)

• FAPESP's process: 19/14582-8 - Automatic adaptation of annotations in the semantic web
• Grantee: Enio de Jesus Pontes Monteiro
• Support Opportunities: Scholarships in Brazil - Master