24
A. Chatzimichail et al.
3.3.6 Handling Noise, Uncertainty and Imperfect Information
As masses of data have grown tremendously during the past years due to, but not
exclusively, the eruption of the IoT field, it only seems logical to have arisen issues
regarding the quality of such data. Towards this direction several state-of-the-art
frameworks have been proposed to address malfunctions deriving from data noise,
data uncertainty and imperfect information [142].
According to [143], a logical reasoning model has been used to predict missing
data. Grounded on ontological domain knowledge along with a satisfactory dataset
of statements, logical reasoning can track inconsistencies and infer new statements
as predictions of missing data.
Fuzzy reasoning, encompassing all the properties defined in fuzzy logic theory
described in [144], and more specifically non-monotonic reasoning is based on the
concept that an assertion can be generated from premises not entirely specified, but
in the occasion of an exception emerging the conclusion can be withdrawn [145].
Unfortunately, experiential research regarding adding non-monotonic layers upon
reasoning to deal with uncertainty and conflicting data is sparse and not systematic
as in the case of [146], where a rule base compression approach is suggested for the
decrease of non-monotonic rules, or in the case of [147] where a framework, called
FUSE, integrating fuzzy reasoning and semantic reasoning was developed towards
a unified reasoning process for the provision of personalized learning recommendations adaptively and semantically. In addition, a proposal presenting fuzzy analogical
reasoning has been conducted where the case study of MiMo incorporating soft computing showcases the evaluation [148].
Finally, at the exertion of tackling the nuisance of imperfect information upon
reasoning, several investigations were completed, such as in alternating-time temporal logic (ATL) about responsibility in multiagent systems [149] or agents with
perfect recall where the past is not forgotten in nested games [150]. Furthermore,
investigations towards Graded Computation Tree Logic with finite path semantics
(GCTL*f) under imperfect information settings were performed [151].
4 The Semantic Web of Things and How It Augments
the IoT
With the advent of IoT hundreds of sensors, smart devices and smartphones have
been deployed in our everyday lives. The result of this is tremendous amounts of
data with great differences in formats and domains. This has posed great challenges
for machines to understand information and extract knowledge from those data. For
better representation of IoT different data research studies have proposed different techniques to enable machines to intelligently understand heterogeneous data.
Semantic Web of Things (SWoT) is a continuation of World Wide Web that tries
to solve the problems arised from the heterogeneous systems and provides a bet-
A. Chatzimichail et al.
3.3.6 Handling Noise, Uncertainty and Imperfect Information
As masses of data have grown tremendously during the past years due to, but not
exclusively, the eruption of the IoT field, it only seems logical to have arisen issues
regarding the quality of such data. Towards this direction several state-of-the-art
frameworks have been proposed to address malfunctions deriving from data noise,
data uncertainty and imperfect information [142].
According to [143], a logical reasoning model has been used to predict missing
data. Grounded on ontological domain knowledge along with a satisfactory dataset
of statements, logical reasoning can track inconsistencies and infer new statements
as predictions of missing data.
Fuzzy reasoning, encompassing all the properties defined in fuzzy logic theory
described in [144], and more specifically non-monotonic reasoning is based on the
concept that an assertion can be generated from premises not entirely specified, but
in the occasion of an exception emerging the conclusion can be withdrawn [145].
Unfortunately, experiential research regarding adding non-monotonic layers upon
reasoning to deal with uncertainty and conflicting data is sparse and not systematic
as in the case of [146], where a rule base compression approach is suggested for the
decrease of non-monotonic rules, or in the case of [147] where a framework, called
FUSE, integrating fuzzy reasoning and semantic reasoning was developed towards
a unified reasoning process for the provision of personalized learning recommendations adaptively and semantically. In addition, a proposal presenting fuzzy analogical
reasoning has been conducted where the case study of MiMo incorporating soft computing showcases the evaluation [148].
Finally, at the exertion of tackling the nuisance of imperfect information upon
reasoning, several investigations were completed, such as in alternating-time temporal logic (ATL) about responsibility in multiagent systems [149] or agents with
perfect recall where the past is not forgotten in nested games [150]. Furthermore,
investigations towards Graded Computation Tree Logic with finite path semantics
(GCTL*f) under imperfect information settings were performed [151].
4 The Semantic Web of Things and How It Augments
the IoT
With the advent of IoT hundreds of sensors, smart devices and smartphones have
been deployed in our everyday lives. The result of this is tremendous amounts of
data with great differences in formats and domains. This has posed great challenges
for machines to understand information and extract knowledge from those data. For
better representation of IoT different data research studies have proposed different techniques to enable machines to intelligently understand heterogeneous data.
Semantic Web of Things (SWoT) is a continuation of World Wide Web that tries
to solve the problems arised from the heterogeneous systems and provides a bet-