PerfectO: An Online Toolkit for Improving Quality …
179
been released and disseminated by ETSI, W3C, AOITI, etc. to guide IoT and standard developers (also from OneM2M and ISO) to reuse and develop semantics-based
IoT applications easily. The FIESTA-IoT ontology [16] has been employed within
the FIESTA-IoT H2020 EU project involving 14 academic and industrial partners, it
enables to process data generated by sensor within smart cities. The LOV4IoT dataset
is used within experiments to rank the popularity of IoT ontologies in [51] and consider the following metrics: availability, believability, understandability, interlinking,
PageRank, consistency, and richness. PerfectO is also refined and applied within the
IEEE Autonomous robotics ontology community [52], and for well-being and affective science with the design of a recommender system for happiness which encourages the design and reuse of ontologies for emotion, food, obesity, fitness, sleep,
stress, depression, acupuncture, etc. [53].
4.4 Limitations
The current demonstrators have some limitations. The current Graphical User Interfaces (GUI) can be more user-friendly. The demonstrators can be extended with
additional tools such as the use of ontology matching tools. The overall PerfectO
demonstrator is hosted on the Google Platform to avoid maintenance and security
server issues. However, latency issues can appear to load the website and query web
services. Sometimes, we notice that some tools are not maintained anymore. However, we regularly update our demonstrators. For instance, Parrot has been duplicated
and hosted on another server by Mondeca Labs.
Ideally, in the future, PerfectO distinguish four types of resources:
• An Ontology is a set of concepts and relationships between concepts to describe a
specific domain. We designed the Ontology Improvement methodology explained
in Sect. 4.2 which is the main focus of this paper.
• A Dataset is structured according to ontologies.
• A Reasoning or inference engine executes a set of logical rules (e.g., IF THEN
ELSE deductive rules) compliant with the ontologies to extract meaningful information from a dataset.
• A Query can be a SPARQL query compliant with the ontology and the dataset,
used to retrieve a subset of the dataset required to design a specific application.
5 Evaluation, Lessons Learned and Discussions
The PerfectO approach is evaluated with a set of ontologies referenced by the
LOV4IoT catalog in Sect. 5.1. Evaluation criteria defined by the Semantic Web
community are reminded in Sect. 5.2. Discussions including lessons learned are
provided in Sect. 5.3.
179
been released and disseminated by ETSI, W3C, AOITI, etc. to guide IoT and standard developers (also from OneM2M and ISO) to reuse and develop semantics-based
IoT applications easily. The FIESTA-IoT ontology [16] has been employed within
the FIESTA-IoT H2020 EU project involving 14 academic and industrial partners, it
enables to process data generated by sensor within smart cities. The LOV4IoT dataset
is used within experiments to rank the popularity of IoT ontologies in [51] and consider the following metrics: availability, believability, understandability, interlinking,
PageRank, consistency, and richness. PerfectO is also refined and applied within the
IEEE Autonomous robotics ontology community [52], and for well-being and affective science with the design of a recommender system for happiness which encourages the design and reuse of ontologies for emotion, food, obesity, fitness, sleep,
stress, depression, acupuncture, etc. [53].
4.4 Limitations
The current demonstrators have some limitations. The current Graphical User Interfaces (GUI) can be more user-friendly. The demonstrators can be extended with
additional tools such as the use of ontology matching tools. The overall PerfectO
demonstrator is hosted on the Google Platform to avoid maintenance and security
server issues. However, latency issues can appear to load the website and query web
services. Sometimes, we notice that some tools are not maintained anymore. However, we regularly update our demonstrators. For instance, Parrot has been duplicated
and hosted on another server by Mondeca Labs.
Ideally, in the future, PerfectO distinguish four types of resources:
• An Ontology is a set of concepts and relationships between concepts to describe a
specific domain. We designed the Ontology Improvement methodology explained
in Sect. 4.2 which is the main focus of this paper.
• A Dataset is structured according to ontologies.
• A Reasoning or inference engine executes a set of logical rules (e.g., IF THEN
ELSE deductive rules) compliant with the ontologies to extract meaningful information from a dataset.
• A Query can be a SPARQL query compliant with the ontology and the dataset,
used to retrieve a subset of the dataset required to design a specific application.
5 Evaluation, Lessons Learned and Discussions
The PerfectO approach is evaluated with a set of ontologies referenced by the
LOV4IoT catalog in Sect. 5.1. Evaluation criteria defined by the Semantic Web
community are reminded in Sect. 5.2. Discussions including lessons learned are
provided in Sect. 5.3.