Two-Sided Matching Under Incomplete Information

Abstract

In many contexts, stakeholders' preferences are exploited in decision-making. Because of its countless applications in business and the huge number of involved questions, such context has received substantial attention in different domains such as economics, political science, philosophy, and in recent years, computer science. Despite a considerable literature body that studied this kind of context, most efforts assume the availability of precise and complete information about the stakeholders' preferences needed by the decision-making process. Nevertheless, this assumption is invalid because of the confidentiality issues and immense cognitive burden. The target of this study is to formally discuss these restrictions by focusing on prior studies that look at dealing with partial information and proposing solution notions and concepts that assist the development of methods and algorithms that work with inaccurate and partial information in multiple contexts. The paper focuses on the decision-making process under partial information. At the begging, the study address informally the following question: under partial information about the stakeholder preferences, how can we develop an algorithm that is ‘good’, in other words, an algorithm that produces “good” results regarding the complete intrinsic preferences. The paper looks at this problem in a modified version of the two-sided matching problem and shows how to design an approximately-powerful algorithm in such contexts.