In today's global environment, effective communication between groups of diverse ideological beliefs can mean the difference between peaceful negotiations and violent conflict. At the root of communication is language, and researchers at the University of Virginia Center for Religion, Politics, and Conflict (RPC) hypothesize that the analysis of the performative character of a group's discourse (how words are used) provides valuable guidance for how to negotiate with groups of differing ideological beliefs. However, high pressure situations leave little time for an exhaustive analysis of this nature. To address this challenge, this paper expands on the signal processing approach of previous work in the literature, which evaluated the efficacy of a computational approach to applying performative analysis to predict linguistic rigidity. Significantly, this paper evaluates the generalizability of computational performative analysis by considering text from non-religious groups. These include groups focused on political and social agendas rather than religion. The key computational and analytical improvements described in this paper include an enriched judgment selection process and the extraction and analysis of pronoun usage. By examining the raw text of various religious and ideological groups, results show an improved accuracy of 97% for predicting linguistic rigidity, compared to the best predictive accuracy of 83% reported in previous work. These results strengthen the evidence for the hypothesis of the effectiveness of computationally implemented performative analysis as predictive of linguistic rigidity. The results also provide evidence that this approach is applicable to non-religious groups since the predictive accuracy is as consistent with these groups as it is for religious groups.
[1]
Karen A. Maitland,et al.
Pronominal selection and ideological conflict
,
1987
.
[2]
R. Appleby,et al.
Strategic Peacebuilding: An Overview
,
2010
.
[3]
Gurpreet Singh Lehal,et al.
A Survey of Text Mining Techniques and Applications
,
2009
.
[4]
Davide Buscaldi,et al.
A Random Forest Approach for Authorship Profiling
,
2015,
CLEF.
[5]
Don A. Moore,et al.
Barriers to Resolution in Ideologically Based Negotiations: The Role of Values and Institutions
,
2001
.
[6]
Evgeniy Gabrilovich,et al.
Computing Semantic Relatedness Using Wikipedia-based Explicit Semantic Analysis
,
2007,
IJCAI.
[7]
Isabel Iñigo-Mora,et al.
On the use of the personal pronoun we in communities
,
2004
.
[8]
Jeffrey Pennington,et al.
GloVe: Global Vectors for Word Representation
,
2014,
EMNLP.
[9]
C. Lee Giles,et al.
Disambiguating authors in academic publications using random forests
,
2009,
JCDL '09.
[10]
Donald E. Brown,et al.
Predicting the tolerance level of religious discourse through computational linguistics
,
2016,
2016 IEEE Systems and Information Engineering Design Symposium (SIEDS).
[11]
Thorsten Joachims,et al.
Text Categorization with Support Vector Machines: Learning with Many Relevant Features
,
1998,
ECML.
[12]
Anne Kao,et al.
Natural Language Processing and Text Mining
,
2006
.
[13]
Levent Özgür,et al.
Text Categorization with Class-Based and Corpus-Based Keyword Selection
,
2005,
ISCIS.