Mono vs Multilingual BERT: A Case Study in Hindi and Marathi Named Entity Recognition

Litake, Onkar; Sabane, Maithili; Patil, Parth; Ranade, Aparna; Joshi, Raviraj

doi:10.48550/arxiv.2203.12907

Cited by 3 publications

(3 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The monolingual Marathi models released in this work have been shown to work better than the currently available alternatives. A study conducted for Marathi named entity recognition highlights the importance of our models and datasets [11]. A similar study for Marathi text classification and specifically hate speech detection was conducted in [18].…”

Section: Impactmentioning

confidence: 98%

L3Cube-MahaNLP: Marathi Natural Language Processing Datasets, Models, and Library

Joshi¹

2022

Preprint

Self Cite

View full text Add to dashboard Cite

Despite being the third most popular language in India, the Marathi language lacks useful NLP resources. Moreover, popular NLP libraries do not have support for the Marathi language. With L3Cube-MahaNLP, we aim to build resources and a library for Marathi natural language processing. We present datasets and models for supervised tasks like sentiment analysis, named entity recognition, and hate speech detection. We have also published a monolingual Marathi corpus for unsupervised language modeling tasks. Overall we present MahaCorpus, MahaSent, MahaNER, and MahaHate datasets and their corresponding MahaBERT models fine-tuned on these datasets. We aim to move ahead of benchmark datasets and prepare useful resources for Marathi. The resources are available at https://github.com/l3cube-pune/MarathiNLP.

show abstract

Section: Impactmentioning

confidence: 98%

L3Cube-MahaNLP: Marathi Natural Language Processing Datasets, Models, and Library

Joshi¹

2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Named entity recognition (NER) for Hindi and Marathi, two low-resource Indian languages model, is introduced by Litake et al 21 . For NER tasks, transformer-based models are frequently employed.…”

Section: Hindi Text Classificationmentioning

confidence: 99%

Hindi Podcast Genre Prediction using Support Vector Classifier

Mahrishi

Jain

Sharma

2023

Preprint

View full text Add to dashboard Cite

India sees a 23% rise in podcast listening after the Covid-19 pandemic. The pandemic and screen fatigue led people to seek out old favourite audio podcasts. Podcast genre classification allows listeners to compile a playlist of their favourite tracks; this also helps podcast streaming services provide recommendations to users based on the genre of the podcasts they enjoy. After the COVID-19 pandemic, the need for educational content in all forms, including podcasts, has skyrocketed, making it even more crucial to anticipate the genre of educational podcasts. Educational podcasts are a sub-genre of the broader education genre and typically involve audio recordings of discussions, lectures, or interviews on educational topics. Education podcast genre prediction is required to efficiently classify and arrange educational content and make it simpler for listeners to access and absorb pertinent information. This study focuses on Podcast Genre Prediction, specifically for the Hindi language. In our study, our developed PodGen dataset is used, which consists of 550 podcasts of 5 minutes each and have a total of 26,867 sentences, where every podcast has been manually annotated into one of the four genre categories (Horror, Motivational, Crime, and Romance). The performance comparison of state-of-the-art machine learning techniques on the PodGen dataset is used to demonstrate accuracy. The best performance was observed in the case of the Support Vector Classifier model with balanced accuracy:82.42%, precision (weighted):83.09%, recall (weighted):82.42%, and F1 score(weighted):82.39% on testing data.

show abstract

“…We evaluate monolingual and multilingual BERT models on Marathi corpus to compare the performance. A similar analysis for Hindi and Marathi named entity recognition has been performed in [13].…”

Section: Introductionmentioning

confidence: 99%

Mono vs Multilingual BERT for Hate Speech Detection and Text Classification: A Case Study in Marathi

Velankar¹,

Patil²,

Joshi³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

Transformers are the most eminent architectures used for a vast range of Natural Language Processing tasks. These models are pre-trained over a large text corpus and are meant to serve state-of-the-art results over tasks like text classification. In this work, we conduct a comparative study between monolingual and multilingual BERT models. We focus on the Marathi language and evaluate the models on the datasets for hate speech detection, sentiment analysis, and simple text classification in Marathi. We use standard multilingual models such as mBERT, indicBERT, and xlm-RoBERTa and compare them with MahaBERT, MahaALBERT, and MahaRoBERTa, the monolingual models for Marathi. We further show that Marathi monolingual models outperform the multilingual BERT variants in five different downstream fine-tuning experiments. We also evaluate sentence embeddings from these models by freezing the BERT encoder layers. We show that monolingual MahaBERT-based models provide rich representations as compared to sentence embeddings from multi-lingual counterparts. However, we observe that these embeddings are not generic enough and do not work well on out-of-domain social media datasets. We consider two Marathi hate speech datasets L3Cube-MahaHate, HASOC-2021, a Marathi sentiment classification dataset L3Cube-MahaSent, and Marathi Headline, Articles classification datasets.

show abstract

Mono vs Multilingual BERT: A Case Study in Hindi and Marathi Named Entity Recognition

Cited by 3 publications

References 0 publications

L3Cube-MahaNLP: Marathi Natural Language Processing Datasets, Models, and Library

L3Cube-MahaNLP: Marathi Natural Language Processing Datasets, Models, and Library

Hindi Podcast Genre Prediction using Support Vector Classifier

Mono vs Multilingual BERT for Hate Speech Detection and Text Classification: A Case Study in Marathi

Contact Info

Product

Resources

About