Non-Contextual Modeling of Sarcasm using a Neural Network Benchmark
📝 Abstract
One of the most crucial components of natural human-robot interaction is artificial intuition and its influence on dialog systems. The intuitive capability that humans have is undeniably extraordinary, and so remains one of the greatest challenges for natural communicative dialogue between humans and robots. In this paper, we introduce a novel probabilistic modeling framework of identifying, classifying and learning features of sarcastic text via training a neural network with human-informed sarcastic benchmarks. This is necessary for establishing a comprehensive sentiment analysis schema that is sensitive to the nuances of sarcasm-ridden text by being trained on linguistic cues. We show that our model provides a good fit for this type of real-world informed data, with potential to achieve as accurate, if not more, than alternatives. Though the implementation and benchmarking is an extensive task, it can be extended via the same method that we present to capture different forms of nuances in communication and making for much more natural and engaging dialogue systems.
💡 Analysis
One of the most crucial components of natural human-robot interaction is artificial intuition and its influence on dialog systems. The intuitive capability that humans have is undeniably extraordinary, and so remains one of the greatest challenges for natural communicative dialogue between humans and robots. In this paper, we introduce a novel probabilistic modeling framework of identifying, classifying and learning features of sarcastic text via training a neural network with human-informed sarcastic benchmarks. This is necessary for establishing a comprehensive sentiment analysis schema that is sensitive to the nuances of sarcasm-ridden text by being trained on linguistic cues. We show that our model provides a good fit for this type of real-world informed data, with potential to achieve as accurate, if not more, than alternatives. Though the implementation and benchmarking is an extensive task, it can be extended via the same method that we present to capture different forms of nuances in communication and making for much more natural and engaging dialogue systems.
📄 Content
Non-Contextual Modeling of Sarcasm using
a Neural Network Benchmark
Vinay Ashokkumar and N. Dianna Radpour Department of Computer Science, Department of Linguistics State University of New York at Buffalo {vinayash, diannara}@buffalo.edu Abstract One of the most crucial components of natural
human-robot interaction is artificial intuition and its
influence on dialog systems. The intuitive capability
that humans have is undeniably extraordinary, and so
remains one of the greatest challenges for natural
communicative dialogue between humans and robots.
In this paper, we introduce a novel probabilistic
modeling framework of identifying, classifying and
learning features of sarcastic text via training a neural
network with human-informed sarcastic benchmarks.
This is necessary for establishing a comprehensive
sentiment analysis schema that is sensitive to the
nuances of sarcasm-ridden text by being trained on
linguistic cues. We show that our model provides a
good fit for this type of real-world informed data, with
potential to achieve as accurate, if not more, than
alternatives. Though the implementation and
benchmarking is an extensive task, it can be extended
via the same method that we present to capture
different forms of nuances in communication and
making for much more natural and engaging dialogue
systems.
Introduction Motivation Sarcasm has become an increasingly observed nuance in
our everyday communication. It primarily exists in the
form of ironic or satirical (Riloff et al. 2013) discussion.
The use of sarcasm can be seen as having evolved and
popularized since the era of online and virtual
communication, with its use becoming more common and
frequent in conversational settings. This can be evidenced
through the study conducted by (Phillips et al. 2015), in
which they demonstrated the prevalence of sarcasm in
conversation among individuals today, that the current
generation far exceeded in ability to identify sarcasm than
the older generation. Hence, the area of sarcasm detection
within the domains of sentiment analysis, human-computer
interaction and opinion mining, is a complicated problem
in natural language processing, where creating the best
techniques to identify people’s opinions expressed in
written language is a great challenge with huge potential,
as stated by (Farhadloo and Rolland 2016) and (Joshi,
Bhattacharyya, and Carman 2016). Feature-based classification of sarcasm has
become one of the prevalent approaches in the realm of
natural language processing research , where a variety of
lexical, semantic and punctual based features have been
tested through a variety of techniques in identifying and
classifying sarcastic intent in text. Sarcastic dialogue has
been known to be expressed through an exhibition of
several written linguistic properties. It is these through
properties (which can also be assumed to be features), that
the context of irony in which sarcasm is expressed, can be
discovered. These features have been analyzed in various
studies that have mined data in repositories such as Twitter
(Gonzalez-Ibanez, Muresan, and Wacholder 2011),
(Filatova 2012) and Amazon (Davidov, Tsur, and
Rappoport 2010b). A point should be noted that in the case of
incomplete data or scenarios, the analysis of context
through which sarcastic communication can be identified is
much harder as with the lack of data and understanding of
situational meaning. Such cases include incomplete
datasets, noisy or tampered data, or mixed language and
partially ciphered text. Therefore, it is useful to gain insight
in the textual features present in text depicting sarcasm, so
as to improve prediction capability of written sarcastic
bodies through the absence of previous context data or in
the presence of noise. Consider the two sarcastic sentences
below: Haha! I’m trying to imagine you with a
person
This content is AI-processed based on ArXiv data.