This document describes a recommender system developed for the CSIRO Data Access Portal to help users discover research data. It examines two types of recommender systems: content-based, which uses item properties, and collaborative filtering, which uses user similarities. The system was built using both explicit metadata and implicit user behavior data. It calculates similarity between datasets across various features using techniques like TF-IDF and develops recommendations. An evaluation found the top recommendations were relevant 98% of the time. Future work includes enhancing the model and further evaluation.