The development of information technology has led to a sharp increase in data volume. The tremendous amount of data has become a strategic capital that allows businesses to derive superior market intelligence or improve existing operations. People expect to consolidate and utilize data as much as possible. However, too much data will bring huge integration cost, such as the cost of purchasing and cleaning. Therefore, under the context of limited resources, obtaining more data integration value is our expectation. In addition, the uneven quality of data sources make the multi-source selection task more difficult, and low-quality data sources can seriously affect integration results without the desired quality gain. In this paper, we have studied how to balance data gain and cost in the source selection, specifically, maximizing the gain of data on the premise of a given budget. We proposed an improved greedy genetic algorithm (IGGA) to solve the problem of source selection, and carried out a wide range of experimental evaluations on the real and synthetic dataset. The empirical results show considerable performance in favor of the proposed algorithm in terms of solution quality.