Home // UBICOMM 2019, The Thirteenth International Conference on Mobile Ubiquitous Computing, Systems, Services and Technologies // View article

Automating the Semantic Labeling of Stream Data

Authors:
Konstantinos Kotis

Keywords: semantic label; ontology; stream data

Abstract:
The collection of a voluminous real-world stream data is achieved today through a large number of distributed and heterogeneous data sources. On the other hand, it is quite rare to discover and collect semantic models associated with this data, in order to be able to represent implicit meaning and specifying related uncovered concepts and relationships between them. Such semantic models, however, are the key to make the data easily available, understandable and interlinkable for its potential users and applications. Manually modeling the semantics of data requires significant effort and expertise. Most of the related work focuses on the semantic labeling/annotation of the data fields (source attributes), given that a semantic model is already provided. Constructing a semantic model that explicitly describes the relationships between the data attributes in addition to their semantic types is critical. Related works support the semantic annotation of data using existing ontologies, but there are only a few that automatically construct the ontology based on the real-world stream data that will eventually annotate (two-step process). More important, existing solutions require a manually-created training data set and its mapping to existing related ontologies/models, in order to assist in the process of learning the mapping function between the actual stream data and the related semantic model (usually via a supervised machine learning approach). This paper a) presents the problem and representative related work, and b) proposes design directions that are aligned to key requirements.

Pages: 55 to 62

Copyright: Copyright (c) IARIA, 2019

Publication date: September 22, 2019

Published in: conference

ISSN: 2308-4278

ISBN: 978-1-61208-736-8

Location: Porto, Portugal

Dates: from September 22, 2019 to September 26, 2019