Hi, I'm Ashish Rao

This paper proposes a unique approach to cluster these documents using the mini batch k-means algorithm on dimensionally reduced sentence embeddings generated with the use of DistilBERT and UMAP. The proposed approach has been compared to state-of-the-art topic modelling and clustering approaches and has outperformed them.

The paper explores how the evolution of data — from structured to predominantly unstructured — has created a need for scalable, flexible storage solutions that traditional relational databases cannot provide. This has led to the rise of NoSQL databases, which are especially suitable for domains like finance that generate large, dynamic datasets daily. The study compares different types of NoSQL databases based on metrics such as data model, indexing, atomicity, and integrity. It also implements and evaluates MongoDB, Cassandra, and Redis using financial data, measuring their performance on various operations including basic and complex READ and WRITE queries, as well as aggregation tasks.

This paper focuses on performing sentiment analysis on chat messages from Twitch livestreams, which are challenging to analyze due to large volume and platform-specific language like emotes and memes. Several machine learning models were tested, including Support Vector Classifier, Logistic Regression, Decision Tree, Random Forest, and Multinomial Naïve Bayes. The Support Vector Classifier achieved the best performance, surpassing the previous state-of-the-art with notable improvements: 10.3% higher accuracy, 9.1% higher recall, and a 7.5% increase in F1-score.