In a first step towards improving Dutch emotion detection, we try to combine the Dutch transformer models BERTje and RobBERT with lexicon-based methods. We propose two architectures: one in which lexicon information is directly injected into the transformer model and a meta-learning approach where predictions from transformers are combined with lexicon features. The models are tested on 1,000 Dutch tweets and 1,000 captions from TV-shows which have been manually annotated with emotion categories and dimensions. We find that RobBERT clearly outperforms BERTje, but that directly adding lexicon information to transformers does not improve performance. In the meta-learning approach, lexicon information does have a positive effect on BERTje, but not on RobBERT. This suggests that more emotional information is already contained within this latter language model.

Seeing the myriad of existing emotion models, with the categorical versus dimensional opposition the most important dividing line, building an emotion-annotated corpus requires some well thought-out strategies concerning framework choice. In our work on automatic emotion detection in Dutch texts, we investigate this problem by means of two case studies. We find that the labels joy, love, anger, sadness and fear are well-suited to annotate texts coming from various domains and topics, but that the connotation of the labels strongly depends on the origin of the texts. Moreover, it seems that information is lost when an emotional state is forcedly classified in a limited set of categories, indicating that a bi-representational format is desirable when creating an emotion corpus.

This paper presents an emotion classification system for English tweets, submitted for the SemEval shared task on Affect in Tweets, subtask 5: Detecting Emotions. The system combines lexicon, n-gram, style, syntactic and semantic features. For this multi-class multi-label problem, we created a classifier chain. This is an ensemble of eleven binary classifiers, one for each possible emotion category, where each model gets the predictions of the preceding models as additional features. The predicted labels are combined to get a multi-label representation of the predictions. Our system was ranked eleventh among thirty five participating teams, with a Jaccard accuracy of 52.0% and macro- and micro-average F1-scores of 49.3% and 64.0%, respectively.