Dealing with Evaluative Expressions and Hate Speech Metaphors with Fuzzy Property Grammar Systems

We introduce a Fuzzy Property Grammar System (FPGS), a formalism that integrates a Fuzzy Property Grammar into a linguistic grammar system to formally characterize metaphorical evaluative expressions. The main scope of this paper is to present the formalism of FPGS and to show how it might provide a...

Full description

Saved in:
Bibliographic Details
Published in:Axioms 2023-05, Vol.12 (5), p.484
Main Authors: Torrens-Urrutia, Adrià, Jiménez-López, Maria Dolores, Campillo-Muñoz, Susana
Format: Article
Language:English
Subjects:
Context
Cooperation
evaluative expressions
Formalism
Fuzzy algorithms
fuzzy grammar
Fuzzy logic
Fuzzy systems
Grammar
grammar systems
Hate speech
Language
Linguistics
Machine learning
Metaphor
metaphors
Sentiment analysis
Social networks
System theory
vagueness
Violence
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We introduce a Fuzzy Property Grammar System (FPGS), a formalism that integrates a Fuzzy Property Grammar into a linguistic grammar system to formally characterize metaphorical evaluative expressions. The main scope of this paper is to present the formalism of FPGS and to show how it might provide a formal characterization of hate speech linguistic evaluative expressions with metaphors (as fuzzy concepts), together with evaluating their degree of linguistic violence. Linguistic metaphors are full of semantic coercions. It is necessary to formally characterize the context of the communication to acknowledge the extralinguistic constraints of the pragmatic domain, which establishes whether an utterance is violent. To show the applicability of our formalism, we present a proof of concept. By compiling and tagging a 3000-tweet corpus, we have extracted a lexicon of hate speech metaphors. Furthermore, we show how FPGS architecture can deal with different types of hate speech and can identify implicit violent figurative evaluative expressions by context and type. Although we are still in the experimental phase of our project and cannot present conclusive results at the computational level, the proof-of-concept results show that our formalism can achieve the desired outcome.
ISSN:2075-1680
2075-1680
DOI:10.3390/axioms12050484