Distributed Computing and Internet Technology

Semantic Web technology has established a framework for creating a “web of data” where the nodes correspond to resources of interest in a domain and the edges correspond to logical statements that link these resources using binary relations of interest in the domain. The framework provides a standardized way of describing a domain of interest so that the description is machine-processable. This enables applications to share data and knowledge about entities in an unambiguous manner. Also, as all resources are represented using IRIs, a massive distributed network of datasets gets created. Applications can dynamically discover these datasets, access most recent data, interpret it using the associated meta-data (ontologies) and integrate them into their operations. While the Linked Open Data (LOD) initiative, based on the Semantic Web standards, has resulted in a huge web corpus of domain datasets, it is well-known that the majority of the statements in a dataset are of the type that link specific individuals to specific individuals (e.g. Paris is the capital of France) and there is major need to augment the datasets with statements that link higher-level entities (e.g. A statement about Countries and Cities such as “Every country has a city as its capital”). Adding statements of this kind is part of the task of enrichment of the LOD datasets called “ontology enrichment”. In this paper, we review various recent research efforts that address this task. We investigate different types of ontology enrichments that are possible and summarize the research efforts in each category. We observe that while the initial rapid growth of LOD was contributed by techniques that converted structured data into the LOD space, the ontology enrichment is more involved and requires several techniques from natural language processing, machine learning and also methods that cleverly make use of the existing ontology statements to obtain new statements.