Inspiration
As an AI enthusiast and a social media user, I was inspired to create a project that could harness the power of Natural Language Processing to understand the sentiments expressed in tweets. I wanted to delve into the world of social media and gain insights into the collective emotions of users.
What it does
The Cyberbullying Tweet Recognition Project using Supervised Machine learning and Natural Language Processing is a valuable tool that can be used to identify and prevent cyberbullying on social media platforms. The project uses advanced machine learning algorithms and natural language processing techniques to analyze social media data and detect instances of cyberbullying.
How we built it
We have used SVC (Support Vector Classifier) algorithm to implement the project after comparing different models like logistic regression, Naive Bayes, decision tree, random forest, adaboost classifiers , where we have processed tweet/twitter (text) dataset and have trained our model.
Challenges we ran into
We have faced challenges regarding the selection and accuracy of the model, we have improved our model accuracy by making changes in preprocessing of text and other functions.
- Data Quality: Obtaining a high-quality and balanced dataset was a challenge. Cleaning and annotating tweets can be time-consuming.
- Model Fine-Tuning: Achieving a high level of accuracy required extensive fine-tuning of the models, as well as handling issues like class imbalance.
- Real-Time Analysis: Implementing real-time sentiment analysis and hosting the web interface came with its share of technical challenges.
Accomplishments that we're proud of
We are very proud that after a lot of hard work and time we have managed to increase are validation accuracy to about 83%.
What we learned
Throughout this project, I learned the nuances of sentiment analysis, a branch of NLP. I discovered how to process text data, perform tokenization, and apply machine learning models to determine sentiment polarity (positive, negative, neutral). Additionally, I gained experience in utilizing pre-trained language models, such as BERT and GPT-3, for more advanced sentiment analysis.
What's next for Tweet Sentiment Recogniser
The future scope for the Tweet Sentiment Analysis project includes multilingual support, real-time social media monitoring, topic-based analysis, and emotion recognition. It can also be enhanced with advanced deep learning models, customized sentiment analysis for businesses, and interactive dashboards for users. Ensuring accuracy, addressing ethical concerns, and expanding to provide user engagement metrics are key considerations for future development.
Log in or sign up for Devpost to join the conversation.