PT Unknown
AU Abdel-Qader, M
TI Vocabulary Evolution on the Semantic Web: From Changes to Evolution of Vocabularies and its Impact on the Data
PY 2021
PU Christian-Albrechts-Universität zu Kiel
WP https://macau.uni-kiel.de/receive/macau_mods_00001111
LA en
DE Semantic web; Linked Data; Vocabularies; rdf; owl; triples; linked vocabularies; DyLDO; BTC; wikidata; lov
AB The main objective of the Semantic Web is to provide data on the web well-defined meaning. Vocabularies are used for modeling data in the web, provide a shared understanding of a domain and consist of a collection of types and properties. These types and properties are so-called terms. A vocabulary can import terms from other vocabularies, and data publishers use vocabulary terms for modeling data. Importing terms via vocabularies results in a Network of Linked vOcabularies (NeLO). Vocabularies are subject to change during their lifetime. When vocabularies change, the published data become a problem if they are not updated based on these changes. So far, there has been no study that analyzes vocabulary changes over time. Furthermore, it is unknown how data publishers reflect on such vocabulary changes. Ontology engineers and data publishers may not be aware of the changes in the vocabulary terms that have already happened since they occur rather rarely.

This work addresses the problem of vocabulary changes and their impact on other vocabularies and the published data. We analyzed the changes of vocabularies and their reuse. We selected the most dominant vocabularies, based on their use by data publishers. Additionally, we analyzed the changes of 994 vocabularies. Furthermore, we analyzed various vocabularies to better understand by whom and how they are used in the modeled data, and how these changes are adopted in the Linked Open Data cloud. We computed the state of the NeLO from the available versions of vocabularies for over 17 years. We analyzed the static parameters of the NeLO such as its size, density, average degree, and the most important vocabularies at certain points in time. We further investigated how NeLO changes over time, specifically measuring the impact of a change in one vocabulary on others, how the reuse of terms changes, and the importance of vocabulary changes.

Our results show that the vocabularies are highly static, and many of the changes occurred in annotation properties. Additionally, 16% of the existing terms are reused by other vocabularies, and some of the deprecated and deleted terms are still reused. Furthermore, most of the newly coined terms are adopted immediately. Our results show that even if the change frequency of terms is rather low, it can have a high impact on the data due to a large amount of data on the web. Moreover, due to a large number of vocabularies in the NeLO, and therefore the increase of available terms, the percentage of imported terms compared with the available ones has decreased over time. Additionally, based on the scores of the average number of exports for the vocabularies in the NeLO, some vocabularies have become more popular over time. Overall, understanding the evolution of vocabulary terms is important for ontology engineers and data publishers to avoid wrong assumptions about the data published on the web. Furthermore, it may foster a better understanding of the impact of the changes in vocabularies and how they are adopted to possibly learn from previous experience. Our results provide for the first time in-depth insights into the structure and evolution of the NeLO. Supported by proper tools exploiting the analysis of this thesis, it may help ontology engineers to identify data modeling shortcomings and assess the dependencies implied by the reusing of a specific vocabulary.
PI Kiel
ER