In a Kaggle competition, I took on the challenge of combating toxic and insincere content on Quora, a platform dedicated to sharing knowledge and insights. Insincere questions, which often propagate false information or promote divisive agendas, pose a significant issue for maintaining a safe and respectful online environment. My objective was to develop machine learning models that could effectively identify and flag such insincere questions.
Using a combination of GLOVE and Paragram embeddings, I constructed a model capable of achieving a commendable 68% F1 score on the private leaderboard. This accomplishment reflects my proficiency in natural language processing and text classification. By participating in this competition, I contributed to Quora’s ongoing efforts to create a more inclusive and respectful online community, upholding their ‘Be Nice, Be Respectful’ policy.
This project in my portfolio highlights my commitment to addressing real-world challenges in the online space, emphasizing my expertise in text-based machine learning and my dedication to fostering a positive digital environment.”
Notebook Link: https://www.kaggle.com/code/dipta007/glove-para/notebook