EmojiText: An Information Visualization Technique for Analyzing Phrases and Sentiments

This paper presents a proposal for text visualization with a focus on the representation of sentences and sentiments. It is noteworthy that traditional text visualization techniques, such as wordcloud, do not visually present relations between words, making it challenging to analyze textual data more deeply through visualization. EmojiText uses a directed graph, where the nodes represent the words and the edges represent the sequence in which the words appear in the sentences. Furthermore, the area of each node is related to the frequency of the represented word, and each node is associated with an emoji that represents the most frequent sentiment associated with the word. Finally, the edges also allow identifying the sentiment at the sentence level. The tool also includes zoom interactions, filters, details on demand, visual mapping configuration, and temporal analysis functionality through animations and multiple views for comparisons. The tool was developed in a javascript programming language using the d3.js framework and is available at http://emojitext.roberto.eti.br/.