Assamese-English Code Mixed dataset with Emotion Labelling using Plutchik’s model

Despite the growing sophistication in multilingual natural language processing (NLP), freely available datasets on emotions annotated for code-mixed text have not been extensively explored yet. Assamese is a low resource langauge and finding pure code mixed data is a trivial task.The dataset on Assamese-English code mixed data serves as a balanced code-mixed corpus of Assamese and English for multi-class emotion recognition in a low-resource multilingual environment.

The dataset features linguistic patterns at the token level, code-mixing statistics, structural analysis using part-of-speech tagging, and emotion labeling, which aid in the study of computational linguistics. The data set can be repurposed for benchmarking machine learning, deep learning, and transformers-based multilingual natural language processing models. This dataset is useful for studies on code-mixing analysis, emotion detection, processing of low-resource languages, and cross-lingual representation learning.