Digital transformation has driven think tank research toward data-driven approaches, by which the quality and effectiveness of information services based on data resources help enhance the scientificity and comprehensiveness of think tank research. This paper classifies the external information required for think tank research into two categories of data resources: open-source and purchased. It systematically reviews the characteristics and constituent elements of these two categories of data resources, establishes a three-dimensional secondary index system (content, carrier and utility) for open-source data resources, and develops a five-dimensional secondary index system (content, function, service, economy and usage) for purchased data resources. Additionally, a model for applying the data resource evaluation system is constructed based on the fuzzy comprehensive evaluation method, providing a foundation for procedural automatic evaluation.