Towards an Interactive Learn-to-Rank System for Economic Competitiveness Understanding

Ranking models are useful tools o‰en employed to aid in decision making. In €elds such as economics, the development of indicators to rank economies or regions are typically dictated by expert opinion. With the increased availability of high €delity open data, beŠer tools for developing and understanding rankings can provide valuable insight into social and economic questions. Œis paper presents a preliminary foray into the development of such tools. We introduce a vision for leveraging state-of-the-art algorithms from the Information Retrieval €eld to design interactive learn-to-rank tools. Incorporated into data analytics systems via plug-and-play components, such tools hold the potential to beŠer evaluate the comparative merits of di‚erent regions and to interactively assess the impact of di‚erent features on the €nal rankings. ŒeMyRanker paradigm is applied in the context of MATTERS, a public system for evaluating the economic competitiveness of US states. A preliminary analysis and discussion of the system highlight its promise for ranking analysis.