A method for automatic estimation of meaning ambiguity of emoticons based on their linguistic expressibility. (January 2020)

• Record Type: Journal Article
• Authors: Ptaszynski, Michal; Masui, Fumito; Ishii, Naoto
• Is Part Of: Cognitive systems research. Volume 59(2020)
• Page Start: 103
• Page End: 113
• Publication Date: 2020-01
• Subjects: Emoticons -- Linguistic expressibility -- Meaning ambiguity; Cognition -- Periodicals; Cognitive engineering (System design) -- Periodicals; Artificial intelligence -- Periodicals; 153.05
• DOI: 10.1016/j.cogsys.2019.09.022
• Languages: English
• ISSNs: 1389-0417

Abstract

Abstract: The ways of how people communicate on the Internet have adapted to the limited communication channel. Since the most meaning-rich while also concise and compressible mean of communication is still represented by textual information, Internet communication has adapted to support it with other information required for natural communication, while retaining the consistency of the medium. One of the means widely used by many generations of Internet users are the emoticons, used to express information that cannot be fully transmitted only by text, such as emotions or feelings. However, not all emoticons developed and widely used everyday are easily understandable to all users. Emoticon usage proficiency differs among specific persons, as well as between more general user groups. Moreover, the meaning of emoticons themselves can also be ambiguous, especially, when used without other information. In this research, we aimed to determine a correspondence between emoticons and linguistic expressions corresponding to them, such as onomatopoeia, etc., to estimate the general level of meaning ambiguity of emoticons. In the research, we first performed a survey in which we asked users about which emoticon conveys which meaning. Then we applied the results of the survey to quantify the understandability, or meaning ambiguity, of the emoticons. Based on the quantification, we propsed a method for specifying the emoticon's meaning ambiguity, in which we applied the aility of the emotion to be expressed in everyday vocabulary, or its linguistic expressibility . Three versions of the proposed method showed that it is possible to estimate emoticon meaning ambiguity, although future experiments and additional data collection is required to maximize the performance. The method could help users apply more meaningful and understandable emoticons in their everyday communication.