DravidianMultiModality: A Dataset for Multi-modal Sentiment Analysis in Tamil and Malayalam

Chakravarthi, Bharathi Raja; P., K, Jishnu Parameswaran; Premjith, B,; Soman, K. P; Ponnusamy, Rahul; Kumaresan, Prasanna Kumar; Thamburaj, Kingston Pal; McCrae, John P.

doi:10.48550/arxiv.2106.04853

Cited by 1 publication

(2 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…HASOC-Dravidian-CodeMix -FIRE 2020 [27] [28] is the first shared task for identifying offensive content in Tamili languages. Previous work on Tamili languages on hope speech [29,30], troll meme detection [31], multimodal sentiment analysis [9] have paved the way to research in Tamili languages.…”

Section: Related Workmentioning

confidence: 99%

“…Though there is a substantial amount of work done on major languages like English to identify offensive content [2], it is a challenging task to identify and flag offensive content in low resource languages, since many users tend to write their language in English script, which is called code-switching or code-mixing [3,4,5]. Developing NLP systems on code-mixed text is challenging since the number of datasets is scarce [6,7,8,9] and there are no clear patterns on these texts. The spelling and context might vary depending upon the user.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Pretrained Transformers for Offensive Language Identification in Tanglish

Benhur¹,

Sivanraju²

2021

Preprint

View full text Add to dashboard Cite

This paper describes the system submitted to Dravidian-Codemix-HASOC2021: Hate Speech and Offensive Language Identification in Dravidian Languages (Tamil-English and Malayalam-English). This task aims to identify offensive content in code-mixed comments/posts in Dravidian Languages collected from social media. Our approach utilizes pooling the last layers of pretrained transformer multilingual BERT for this task which helped us achieve rank nine on the leaderboard with a weighted average score of 0.61 for the Tamil-English dataset in subtask B. After the task deadline, we sampled the dataset uniformly and used the MuRIL pretrained model, which helped us achieve a weighted average score of 0.67, the top score in the leaderboard. Furthermore, our approach to utilizing the pretrained models helps reuse our models for the same task with a different dataset. Our code and models are available in GitHub 1

show abstract

Section: Related Workmentioning

confidence: 99%