The goal of the research was to make processing recursive joins on big datasets in Spark faster and more efficient than in MapReduce. They found that Spark's in-memory computing and caching features could help solve the problems faced in recursive join operations that were complex and expensive in MapReduce. By cutting down on redundant data and using memory smartly, the researchers made significant performance improvements in handling recursive joins with large datasets. Their solution allows for quicker and more effective processing of these complex operations, making data analysis on a large scale smoother and more cost-effective.