Reducing Market Noise: Developing an Open-Source NLP Algorithm for Financial Feed Filtering and Sentiment Personalization

Hello everyone. I am currently working on an open-source natural language processing (NLP) algorithm designed to mitigate informational noise in financial media. The primary objective is to clean RSS and news feeds from pervasive clickbait, speculative shock content, and irrelevant geopolitical noise (such as non-market-moving war topics).The system operates by parsing incoming text, evaluating shock/clickbait probabilities through custom sentiment dictionaries, and structuring the output into clean JSON format. Additionally, it tracks user interaction metrics to dynamically adjust and personalize the feed based on active market interests.I am deploying the alpha version and seeking technical feedback regarding latency optimization and false-positive minimization before integrating the pipeline with algorithmic trading architectures.The GitHub repository link is provided in the comments below. I would greatly appreciate your insights on the methodology.