EmoLabel: Semi-Automatic Methodology for Emotion Annotation of Social Media Text

The exponential growth of the amount subjective information on Web 2.0. has caused an increasing interest from researchers willing to develop methods extract emotion data these new sources. One most important challenges in textual detection is gathering with labels because subjectivity assigning labels. Basing this rationale, main objective our research contribute resolution challenge. This tackled by proposing EmoLabel: a semi-automatic methodology based pre-annotation, which consists two phases: (1) automatic process pre-annotate unlabelled English sentences; and (2) manual refinement where human annotators determine dominant emotion. Our assess influence pre-annotation method annotation points view: agreement time needed for annotation. evaluation performed demonstrates benefits processes since results show gain near 20 percent when applied (Pre-ML) without reducing annotator performance. Moreover, are higher those contributors whose performance low (inaccurate annotators).