Cooking Well, With Love, Is an Art: Transformers on Youtube Hinglish Data

Sargam Yadav, Abhishek Kaushik, Shubham Sharma

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Youtube gives a place for viewers to express their views on a variety of issues. This study focuses on opinion mining from a dataset containing comments from Nisha Madulika's and Kabita's Kitchen's YouTube culinary channels. The information includes Mix code comments in both English and a Hindi-English hybrid known as Hinglish. On both datasets, variations of the Bidirectional Encoder Representations from Transformers (BERT) have been trained. BERT outperformed the baseline trained on machine learning in terms of outcomes. The results show that the innovative BERT models (84.82% and 88.37%) are likely to outperform on Hinglish datasets with fine-tuning.

Original languageEnglish
Title of host publication2021 International Conference on Computational Performance Evaluation, ComPE 2021
EditorsSudip Paul, Jitendra Kumar Verma
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages836-841
Number of pages6
ISBN (Electronic)9781665436564
DOIs
Publication statusPublished - 2021
Event2021 International Conference on Computational Performance Evaluation, ComPE 2021 - Shillong, India
Duration: 1 Dec 20213 Dec 2021

Publication series

Name2021 International Conference on Computational Performance Evaluation, ComPE 2021

Conference

Conference2021 International Conference on Computational Performance Evaluation, ComPE 2021
Country/TerritoryIndia
CityShillong
Period1/12/213/12/21

Keywords

  • BERT
  • Classification
  • Cookery Channels
  • Hinglish
  • Machine Learning
  • Pretrained
  • Sentiment Analysis
  • Trans-formers

Fingerprint

Dive into the research topics of 'Cooking Well, With Love, Is an Art: Transformers on Youtube Hinglish Data'. Together they form a unique fingerprint.

Cite this