Emotion Representation Mapping for Automatic Lexicon Construction (Mostly) Performs on Human Level
This work addresses the proliferation of emotion representation formats in natural language processing, offering a method to construct emotion lexicons automatically and potentially standardize emotion data across languages.
The paper tackled the problem of converting emotion ratings between different representation formats, such as Valence-Arousal-Dominance and Ekman's Basic Emotions, using a new neural network approach. The result showed that the model outperformed previous state-of-the-art methods and produced results nearly as reliable as human annotations, even in cross-lingual settings, with new emotion ratings generated for 13 diverse languages.
Emotion Representation Mapping (ERM) has the goal to convert existing emotion ratings from one representation format into another one, e.g., mapping Valence-Arousal-Dominance annotations for words or sentences into Ekman's Basic Emotions and vice versa. ERM can thus not only be considered as an alternative to Word Emotion Induction (WEI) techniques for automatic emotion lexicon construction but may also help mitigate problems that come from the proliferation of emotion representation formats in recent years. We propose a new neural network approach to ERM that not only outperforms the previous state-of-the-art. Equally important, we present a refined evaluation methodology and gather strong evidence that our model yields results which are (almost) as reliable as human annotations, even in cross-lingual settings. Based on these results we generate new emotion ratings for 13 typologically diverse languages and claim that they have near-gold quality, at least.