Data Linking for the Semantic Web

Abstract

By specifying that published datasets must link to other existing datasets, the 4th linked data principle ensures a Web of data and not just a set of unconnected data islands. The authors propose in this paper the term data linking to name the problem of finding equivalent resources on the Web of linked data. In order to perform data linking, many techniques were developed, finding their roots in statistics, database, natural language processing and graph theory. The authors begin this paper by providing background information and terminological clarifications related to data linking. Then a comprehensive survey over the various techniques available for data linking is provided. These techniques are classified along the three criteria of granularity, type of evidence, and source of the evidence. Finally, the authors survey eleven recent tools performing data linking and we classify them according to the surveyed techniques.

Towards the Multilingual Semantic Web

“…Several activities have to be carried out for cross-lingual interlinking: (1) the selection of relevant and authoritative mono/multilingual datasets to link, (2) the automatic discovery of equivalent and/or related entities between the dataset and the selected external resources, and finally (3) 24 For example, RDF refine (http://refine.deri.ie/) and Apache Any23 (http://any23.apache.org/). There are many tools and techniques for discovering links between data items of different RDF datasets (see Ferrara et al 2011 for a survey). In particular, cross-lingual link discovery involves the automatic discovery of relationships between data items to increase the external connectivity of the RDF dataset in a multilingual scenario.…”

“…), (2) the discovery of links between RDF datasets(Ferrara et al 2011), or ), (2) the discovery of links between RDF datasets(Ferrara et al 2011), or …”

Linked data classification

“…The details about the matching feature(s) that cause similarity need to be stored and they need to be considered during classification. The similarity value is usually proportional to the number of features that two linked data items have in common [6]. This means that two pairs of items can be considered to be equally similar since they have the same number of matching features, also when the sets of matching features are different for the two pairs.…”

“…The field of instance matching, with special focus on LOD, is emerging in the recent years and a reference survey of the main existing approaches is provided in [6]. According to this survey, our matching techniques presented in Section 3 belong to the field of value-oriented techniques and rely on methods for comparison cost reduction based on feature subset selection.…”

Lop

“…preparation and transformation, and also the interlinking process, when decisions about schema matching and data fusion must be registered for future use. Our strategy builds upon previous works [4] [12] where LOD publication is supported by a workflow, defined as a systematic sequence of activities, involving data processing for extraction, cleaning, conforming and pre-integration among different sources (especially if considering publication of datasets belonging to a single organization) previously to a transformation (or triple format conversion) step. We define a novel mechanism that collects provenance data throughout these activities, recording it at various levels of granularity.…”