Page 146 - Handbook of Deep Learning in Biomedical Engineering Techniques and Applications

Chapter 5 Depression discovery in cancer communities using deep learning 135

that learns concepts rather than review documents and
thus decrease the innate vagueness and indistinctness. The exper-
imental results demonstrate the superiority of their proposed
approach across a number of different data sets and ascertained
the capacity of their approach to reduce the sensitivity to clatter
and its flexibility across domains.
Authors in Ref. [56] use ontology-based techniques for SA on
Twitter. Domain ontology helps identify and score the different
aspects in tweets about smartphones. Even though they improve
aspect identification using an ontology, they do not have any po-
larity detection mechanism in place and rely on a web service for
this. Also, to be able to work for other domains, they must have an
automatic ontology building capability.
In Ref. [57], the authors introduce SenticNet 2, an openly avail-
able semantic and emotion-based lexicon for SA. This lexicon
bridges the gap between word-level natural language data and
the concept-level sentiments borne. To build SenticNet 2, the au-
thors exploit sentic computing, a novel hypothesis that exploits
artificial intelligence and semantic Web. They illustrate its work-
ing by embedding in real-world applications for patients' SA [58]
and crowd validation [59]. The use of SenticNet2 is constrained
by the need for techniques and tools that facilitate its seamless
integration with external knowledge bases, to improve the extrac-
tion of semantics and sentics from many different types of media.

2.5Sentiment analysis for online depression
detection
Ascertaining the course of treatment and steps to be followed
to cure a mental disorder is a complex medical assessment, which
should only be carried out by psychiatrists. It encompasses
numerous aspects such as ascertaining the severity of symptoms,
the level of pain and distress initiated by the disease, desirable
and undesirable effects of remedies, incapacities linked with
the symptoms, symptoms that damagingly effect additional ail-
ments, and so on. [2]. Furthermore, establishing the severity of
an ailment is a grueling job that should merely be carried out by
an extremely proficient and certified expert by carrying out one-
to-one interviews and using their judgments. Bearing in mind
the intricacy of the processes and the amount of expertise
required in recognizing and categorizing different mental disor-
ders and their essential treatments, discovering mental ailments
such as depression on social media using SA and affect analysis
is just a first step, which can be used to breed consciousness.