=Paper=
{{Paper
|id=Vol-2721/paper573
|storemode=property
|title=Is your data 6-star?
|pdfUrl=https://ceur-ws.org/Vol-2721/paper573.pdf
|volume=Vol-2721
|authors=Cassia Trojahn
|dblpUrl=https://dblp.org/rec/conf/semweb/Trojahn20
}}
==Is your data 6-star?==
https://ceur-ws.org/Vol-2721/paper573.pdf
Is Your Data 6-Star??
Cassia Trojahn
Institut de Recherche en Informatique de Toulouse, France
firstname.lastname@irit.fr
Abstract. Linked open data in general is poor in foundational distinc-
tions and by consequence lacks semantics clarity. Such distinctions are
however essential for many applications consuming the data, and have
been since many years the subject of study in foundational ontologies.
This paper argues that foundational distinctions have to be better taken
into account in the process of construction, alignment and publication of
linked open data, from a methodological perspective. It proposes to ex-
tend the well-known 5-star rating schema to a 6-star schema, with data
getting a 6-star rating when equipped with foundational distinctions.
1 Introduction
The Berners-Lee’s vision of a semantic web has been materialized with the Linked
Open Data (LOD) initiative, where structured data are (ideally) exposed as in-
stances of ontologies and linked across knowledge bases. In that perspective, the
well-known 5-star incremental framework has guided the process of publishing
data on the Web1 . From there, the LOD cloud has become an extremely rich
source of knowledge in several domains (such as Geography, Linguistics, Life
Sciences, Social Networking) with about 1260 datasets with 16187 links2 .
While most (core) linked open datasets have been constructed from existing
(encyclopedic) sources (such as DBpedia and BabelNet), the process of construc-
tion, alignment and publication of data (and the ontologies describing them)
highly neglects the role of foundational ontologies. In consequence, linked open
data in general is poor in foundational distinctions. Previous works have ad-
dressed, in particular, the challenge of matching domain and foundational on-
tologies [12,13]. More recently, the lack of foundational distinctions in the LOD
has been highlighted in [2]: distinctions such as whether an entity is inherently
a class or an individual, or whether it is a physical object or not, are hardly
expressed in the data, although they have been largely studied and formalised
by foundational ontologies. Such distinctions are essential in many applications
consuming linked open data (and essential in Artificial Intelligence in general).
They are however at the center of the formal and applied ontology field focusing
?
Copyright © 2020 for this paper by its authors. Use permitted under Creative Com-
mons License Attribution 4.0 International (CC BY 4.0).
1
https://www.w3.org/DesignIssues/LinkedData.html
2
https://lod-cloud.net/ (August 2020)
on a large spectrum of foundational issues (types of entities, formal relations,
space, time, etc.). Complementary to [2], the authors in [3] state that in the
semantic web, there is an increasingly need for serious engagement with ontolo-
gies, understood as a general theory of the types of entities and relations making
up their respective domains of inquiry. However, there is still little interaction
between the communities, despite the fact that they share common ambitions
in terms of knowledge understanding.
This paper argues that foundational distinctions have to be better taken
into account in the process of construction, alignment and publication of LOD
data from a methodological point of view. In that perspective, we propose to
extend the 5-star data schema to a 6-star schema, with data getting a 6-star
when equipped with foundational distinctions (as for instance, to be linked to
appropriated foundational ontologies).
2 Which kinds of foundational distinctions?
Foundational distinctions are at the core of foundational (top-level or upper)
ontologies. A foundational ontology is a high-level and domain independent on-
tology whose concepts (e.g., object, event, quality, disposition) and relations
(e.g., parthood, participation, dependence, causality) are intended to be basic
and universal to ensure generality and expressiveness for a wide range of do-
mains. It is often characterized as representing commonsense concepts and is
limited to concepts which are meta, generic, and philosophical. Diverse foun-
dational ontologies have been developed so far (BFO, DOLCE, GFO, SUMO,
UFO, PROTON, to cite a few), influenced by different philosophies and views
on the reality [9,10]. One of the well-known foundational ontologies is DOLCE
[5] (Descriptive Ontology for Linguistic and Cognitive Engineering), an ontology
of particulars which adopts a descriptive approach with a clear cognitive bias, as
it aims at capturing the ontological categories underlying natural language and
human commonsense. DOLCE is based on a fundamental distinction between en-
durant (objects or substances) and perdurant entities (events or processes). The
main relation between endurants and perdurants is that of participation. Under
another perspective, GFO [6] (General Formal Ontology) considers distinctions
between concrete individuals which exist in time or space whereas abstract in-
dividuals do not. While an endurant is an individual that exists in time, but
cannot be described as having temporal parts or phases; a process, on the other
hand, is extended in time. Complementary, BFO [1] (Basic Formal Ontology)
represents the reality into two disjoint categories of continuant (independent and
dependent continuants, attributes, and locations) and occurrent (processes and
temporal regions). A comparison of foundation ontologies can be found in [10].
3 What has been done so far?
From a methodological point of view, LOD highly neglects the role of founda-
tional ontologies. There are two approaches for the use of foundational ontologies
[14]. With a top-down approach, foundational ontologies are used as a reference
for deriving domain concepts, taking advantage of the knowledge and experience
already encoded in it. In a bottom-up approach, one usually matches an existing
domain ontology to the foundational ontology. As reported in [8,9], method-
ologies for constructing ontologies should not neglect the use of foundational
ontologies and should better address it in a top-down approach. In the absence
of systematic adoption of foundational ontologies within the domain ontology
development process (in general), bottom-up approaches have to be applied in-
stead. In this task, matching foundational and domain ontologies plays a key
role. Most state-of-the-art matching systems however fail in the task [13], with
few dedicated approaches been developed so far [7,12].
With respect to aligning and equipping LOD datasets with foundational on-
tologies, in [7], the approach has been used to align the PROTON foundational
ontology to LOD datasets. It uses Wikipedia to construct a set of category
hierarchy trees and then determines which classes to align using different simi-
larities. One proposal analysing the foundational coverage of DBPedia is the one
by [11], where correspondences between DBpedia ontology and DOLCE-Zero
[4], a module of DOLCE, are used to identify inconsistent statements in DBpe-
dia. The authors focus on finding systematic errors or anti-patterns in DBpedia.
They argued that by aligning these ontologies and by combining reasoning and
clustering of the reasoning results, errors affecting statements can be identified
with minimal human workload. More recently, in [2], automatic classification of
foundational distinctions (class vs. instance or physical vs. non-physical objects)
of LOD entities is done with two strategies: an (unsupervised) alignment ap-
proach and a (supervised) machine learning approach. The alignment approach,
in particular, relies on the structure of alignments between DBpedia, DOLCE,
and external lexical linked data. They use the paths of alignments and taxonom-
ical relations in these resources and automated inferences to classifying whether
a DBpedia entity is a physical object or not.
4 The 6-star data rating schema
We propose a 6-star rating schema for data that expresses foundational onto-
logical distinctions. As briefly introduced in Section 2, such distinctions include
clear semantics on, for instance, concrete and abstract individuals, events and
processes, time and temporal regions. They take into account as well formal
relations, such as parthood, dependence, constitution, causality, instantiation.
The proposed rating schema, revising the well-known 5-star schema, is as in the
following:
? Available on the web (whatever format) but with an open licence
?? Available as machine-readable structured data
? ? ? Available with non-proprietary format
? ? ?? All the above plus using open standards from W3C
? ? ? ? ? All the above, plus data linked to other data to provide context
? ? ? ? ?? All the above, plus data equipped with foundational distinctions
5 Final remarks
Foundational distinctions guarantee data consistency and improves semantics
clarity. Linked open data equipped with such distinctions should be rewarded
with a 6-star. In the future, we plan to extend our previous work [12] in order
to take into account data (instance) matching. Complementary to what has
been proposed in [2], such approach could be then applied for helping improving
existing datasets with foundational distinctions.
References
1. R. Arp, B. Smith, and A. Spear. Building Ontologies with Basic Formal Ontology.
MIT Press, 2015.
2. L. Asprino, V. Basile, P. Ciancarini, and V. Presutti. Empirical analysis of foun-
dational distinctions in linked open data. In Proceedings of the Twenty-Seventh
International Joint Conference on Artificial Intelligence, IJCAI 2018, July 13-19,
2018, Stockholm, Sweden., pages 3962–3969, 2018.
3. M. Bennett and K. Baclawski. The role of ontologies in linked data, big data and
semantic web applications. Applied Ontology, 12(3-4):189–194, 2017.
4. A. Gangemi, N. Guarino, C. Masolo, and A. Oltramari. Sweetening WORDNET
with DOLCE. AI Magazine, 24(3):13–24, 2003.
5. A. Gangemi, N. Guarino, C. Masolo, A. Oltramari, and L. Schneider. Sweeten-
ing Ontologies with DOLCE. In 13th Conference on Knowledge Engineering and
Knowledge Management, pages 166–181, 2002.
6. H. Herre, B. Heller, P. Burek, R. Hoehndorf, F. Loebe, and H. Michalek. Gen-
eral Formal Ontology (GFO): A Foundational Ontology Integrating Objects and
Processes. In Basic Principles, Research Group Ontologies in Medicine, 2007.
7. P. Jain, P. Yeh, K. Verma, R. Vasquez, M. Damova, P. Hitzler, and A. Sheth.
Contextual Ontology Alignment of LOD with an Upper Ontology: A Case Study
with PROTON. In Proceedings of the 8th Extended Semantic Web Conference,
pages 80–92, 2011.
8. C. Keet. The use of foundational ontologies in ontology development: An empirical
assessment. In Proceedings of the 8th Extended Semantic Web Conference, pages
321–335, 2011.
9. Z. Khan and C. Keet. ONSET: Automated Foundational Ontology Selection and
Explanation. In Proceedings of the 18th International Conference on Knowledge
Engineering and Knowledge Management, pages 237–251, 2012.
10. V. Mascardi, V. Cordı̀, and P. Rosso. A Comparison of Upper Ontologies. In 8th
AI*IA/TABOO Workshop on Agents and Industry, pages 55–64, 2007.
11. H. Paulheim and A. Gangemi. Serving dbpedia with dolce – more than just adding
a cherry on top. In ISWC 2015, pages 180–196, 2015.
12. D. Schmidt, R. Basso, C. Trojahn, and R. Vieira. Matching domain and top-level
ontologies exploring word sense disambiguation and word embedding. In Emerging
Topics in Semantic Tech., pages 27–38, 2018.
13. D. Schmidt, C. Trojahn, and R. Vieira. Analysing Top-level and Domain Ontol-
ogy Alignments from Matching Systems. In Proc. of the Workshop on Ontology
Matching, pages 1–12, 2016.
14. S. Semy, M. Pulvermacher, and L. Obrst. Toward the use of an upper ontology
for U.S. government and U.S. military domains: An evaluation. Technical report,
MTR 04B0000063, The MITRE Corporation, 2004.