Scribd
Upload
19 views2 pages

Main Project ML Abstract

This document describes a hybrid machine learning system for classifying user-generated text by sentiment and emotion. It consists of two models: an Emotion Classification Model using the GoEmotions dataset with a fine-tuned BERT architecture, and a Sentiment Analysis Model utilizing the Sentiment140 dataset with a Logistic Regression classifier. The system incorporates a custom Gen-Z Slang Dictionary to effectively handle modern slang, making it applicable for various uses such as mental health monitoring and social media analytics.

Uploaded by

shaanghosh609
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
19 views2 pages

Main Project ML Abstract

This document describes a hybrid machine learning system for classifying user-generated text by sentiment and emotion. It consists of two models: an Emotion Classification Model using the GoEmotions dataset with a fine-tuned BERT architecture, and a Sentiment Analysis Model utilizing the Sentiment140 dataset with a Logistic Regression classifier. The system incorporates a custom Gen-Z Slang Dictionary to effectively handle modern slang, making it applicable for various uses such as mental health monitoring and social media analytics.

Uploaded by

shaanghosh609
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd

ABSTRACT

Title: A Hybrid Machine Learning System


for Sentiment and Emotion Classification
of User Text
This project presents a hybrid machine learning system designed to
automatically analyze user-generated text and classify it based on
sentiment (positive/negative) and emotion categories (such as joy,
anger, sadness, fear, etc.). The objective of this system is to build a more
context-aware and human-like text analysis model that can understand
modern internet language, informal expressions, and emotional patterns.

The system consists of two independent models:

1. Emotion Classification Model:


This model is trained using the GoEmotions dataset, a large, human-
annotated dataset developed by Google containing 58,000 text
samples across 27 emotion classes. The model uses a fine-tuned
BERT transformer architecture, which enables high-performance
contextual understanding. BERT’s bidirectional attention mechanism
helps the system detect subtle emotional cues within short sentences.

2. Sentiment Analysis Model:


This model uses the Sentiment140 dataset, containing 1.6 million
tweets labeled as positive or negative. Text is vectorized using the TF-
IDF method, and a Logistic Regression classifier is trained for
efficient large-scale sentiment prediction. This classical approach is
computationally lightweight and extremely effective for large datasets.

To enhance modern slang handling, a custom Gen-Z Slang Dictionary is


integrated into preprocessing to correctly map slang terms (e.g., “rizz”,
“slay”, “no cap”) to their appropriate sentiment or emotion labels.

Together, both models create a robust system capable of understanding


emotional tone and sentiment polarity in real-world text inputs. The output
can be used in applications such as mental-health monitoring, social media
analytics, customer feedback interpretation, chatbots, and content
moderation.

You might also like