Scribd
Upload
36 views3 pages

NLP-Project Proposal

Uploaded by

Muhammad Aman
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
36 views3 pages

NLP-Project Proposal

Uploaded by

Muhammad Aman
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

NLP PROJECT PROPOSAL

Tweet Sentiment Analysis (Positive,


Negative, Neutral)

GROUP MEMBERS
MOHAMMAD – B20102070
SYED MUJTABA - B20102165
Objective:
This project aims to classify tweets into three sentiment categories: Positive, Negative, or
Neutral. This task will be achieved using two different machine learning models: one based on
based on transfer learning or fine tuning of BERT and another based on Large Language Models
(LLMs), potentially utilizing zero-shot or few-shot learning.

Model 1: Transfer Learning with BERT


- Approach: Fine-tune a pre-trained BERT model on a dataset of tweets to understand their
sentiment. BERT’s contextual understanding makes it ideal for nuanced text classifications like
tweets, which may contain slang, abbreviations, or sarcasm.

- Dataset: A publicly available tweet sentiment dataset on Kaggle will be used.


Link to dataset: https://www.kaggle.com/datasets/jp797498e/twitter-entity-sentiment-analysis?resource=download

- Tasks Fine-tune the BERT model using this dataset and optimize for accurate sentiment
classification.

Model 2: Large Language Model (LLM) with Zero/Few-Shot Learning


- Approach: Use an LLM, such as GPT-3 or GPT-4, to perform zero-shot or few-shot learning for
tweet sentiment classification. This model will not require much domain-specific data and can
generalize across various contexts.

- Tasks: Apply zero-shot classification by prompting the model with examples or fine-tune it if a
large dataset is available.

Retrieval Augmented Generation (RAG) Integration:


- Approach: RAG can be used to enhance classification by retrieving relevant information from
an external knowledge base to aid the classification process. For instance, RAG could retrieve
relevant background on trending topics or common tweet patterns for better classification
accuracy.

Team Structure:
- Two-person team: One member focuses on BERT fine-tuning and the other on LLM and RAG
integration. Collaboration will ensure consistency across models.

Deliverables:
1. A fully implemented tweet classification system with two models.
2. Comparative analysis of both models' performance.
3. Documentation and code for each approach, including RAG integration.
Example of data set:

9198 Nvidia Negative Nvidia did delay by 3070 2 weeks.


9198 Nvidia Negative Nvidia really delayed the 3070 several weeks.
9198 Nvidia Negative Nvidia really only delayed the 3070 2 flight week
9198 Nvidia Negative Nvidia really delayed the next 2 weeks.
9193 Nvidia Positive Love EVERYTHING about it.
9193 Nvidia Positive Love EVERYTHING about it �
9193 Nvidia Positive 22 Love EVERYTHING All about it.
9193 Nvidia Positive Love EVERYTHING about anything.
9192 Nvidia Neutral Save $500 on the AERO 15 OLED - Award winning design, stunning 4K OLED panel,
color accurate right out of the box, with i7 core and Nvidia GTX GPU. This is a sweet
work from home, create anywhere laptop! buff.ly/3kd6oRQ
9192 Nvidia Neutral Save $500 on the AERO 15 OLED - Award-winning design, stunning 4K OLED panel,
color-accurate in-stock, with i7 core and Nvidia GTX GPU. This is a sweet work from
home, create anywhere laptop! buff.ly / 3kd6oRQ
9192 Nvidia Neutral Save $500 on the AERO 15 OR - award-winning design, stunning 4K OR panel, color
accuracy straight out of the box, i7 core, and Nvidia X. It's sweet work from home,
create a laptop anywhere! buff.ly / 3kd6oRQ
9192 Nvidia Neutral Save $500 buying the AERO 15 OLED - Award winning design, stunning 4K OLED
panel, color accurate right out of the scratch, Snapdragon i7 core and Nvidia GTX
GPU. This is one sweet work<unk> home, play anywhere laptop! buff.ly/3kd6oRQ

You might also like