Projects / Computer Science / Computer Science project 7

Hinglish Sentiment Analysis of Product Reviews

Approx. component budget: Rs. 0 - 1,000

Objective

Handle spelling variation and code-mixed text, and compare classic models with a multilingual transformer.

What you will build

A model that reads reviews written in mixed Hindi and English (Roman script) and labels them positive, negative or neutral, with a simple web demo.

Results you can expect

• Macro F1 for 3 models
• Effect of text normalisation (for example 'accha', 'acha', 'achha')
• Error analysis on sarcasm
• And a small test set of 100 reviews collected by your team

What you will learn

Code-mixed language problems, tokenisation, transliteration, multilingual models, and building and labelling a dataset.

Tools and notes

Python, scikit-learn, Hugging Face (multilingual BERT or IndicBERT), a public Hinglish dataset or your own labelled reviews.

Resume points from this project

• Built and tested Hinglish Sentiment Analysis of Product Reviews; objective: handle spelling variation and code-mixed text, and compare classic models with a multilingual transformer.
• Measured and analysed macro F1 for 3 models and effect of text normalisation (for example 'accha', 'acha', 'achha'); wrote the report and presented the results in a viva. Add your own numbers: [x].

Replace [x] with your own measured results. Write only what you really did.

More computer science projects

Back to all projects