EICA Team at SemEval-2018 Task 2: Semantic and Metadata-based Features for Multilingual Emoji Prediction

Proceedings of the 12th International Workshop on Semantic Evaluation (SemEval-2018), pages 419–422 New Orleans, Louisiana, June 5–6, 2018. ©2018 Association for Computational Linguistics EICA Team at SemEval-2018 Task 2: Semantic and Metadata-based Features for Multilingual Emoji Prediction Yufei Xie, Qingqing Song East China Normal University, Shanghai, P.R.China yufeixie@ica.stc.sh.cn Abstract The advent of social media has brought along a novel way of communication where meaning is composed by combining short text messages and visual enhancements, the so-called emo- jis. We describe our system for participating in SemEval-2018 Task 2 on Multilingual Emoji Prediction. Our approach relies on combining a rich set of various types of features: seman- tic and metadata. The most important types turned out to be the metadata feature. In sub- task 1: Emoji Prediction in English, our pri- mary submission obtain a MAP of 16.45, Pre- cision of 31.557, Recall of 16.771 and Accu- racy of 30.992. 1 Introduction Emojis are ideograms which are naturally com- bined with plain text to visually complement or condense the meaning of a message (Barbieri et al., 2017). Despite being widely used in social media, their underlying semantics have received little attention from a Natural Language Process- ing standpoint. Barbieri et al. (2016) compare the meaning and usage of emojis across two Span- ish cities: Barcelona and Madrid. Ljubeˇsi´c et al. (2017) present a set of experiments and analyses on predicting the gender of Twitter users based on languageindependent features extracted either from the text or the metadata of users tweets. Miller et al. (2016) performed an evaluation asking human annotators the meaning of emojis, and the sentiment they evoke. People do not al- ways have the same understanding of emojis, in- deed, there seems to exist multiple interpretations of their meaning beyond their designers intent or the physical object they evoke1. Their main con- clusion was that emojis can lead to misunderstand- ings. The ambiguity of emojis raises an interest- ing question in human-computer interaction: how can we teach an artificial agent to correctly inter- pret and recognise emojis use in spontaneous con- versation? The main motivation of our research is that an artificial intelligence system that is able to predict emojis could contribute to better natu- ral language understanding (Novak et al., 2015) and thus to different natural language processing tasks such as generating emoji-enriched social me- dia content, enhance emotion/sentiment analysis systems, and improve retrieval of social network material. 2 Our Approach 2.1 Features We use several semantic features and metadata features to represent the twitter. 2.1.1 Semantic Features Semantic features represent the basic concep- tual components of meaning for any lexical item (Fromkin et al., 2018). An individual semantic feature constitutes one component of a word’s in- tension, which is the inherent sense or concept evoked (O’Grady et al., 1997). Semantic Word Embeddings. We use semantic word embeddings obtained from Word2vec mod- els for GoogleNews. For each twitter, we con- struct the centroid vector from the vectors of all words in that text. centroid(w1..n) = �n i=1 wi …

Content truncated to 500 words