Crowdster is a framework that enables social navigation in web-based visualizations using crowdsourced evaluation. It collects previous users' interaction data from a visualization and integrates it back to help predict trends and evaluate the visualization. Crowdster addresses performance issues like stability and scalability. While in-field evaluation provides immediate summarized results, it risks misinterpreting user behaviors or collecting incorrect data. The framework could potentially be extended to other applications like collecting data from one visualization to feed another or aggregating data across multiple visualizations for group evaluation.