A decision procedure for bundle purchasing with incomplete information on future prices

This paper introduces the on-line bundle-purchasing problem (OBPP) as a new computational challenge induced by e-commerce technology. The task of the OBPP is to decide which of many satisfactory combinations (bundles) of items should be purchased, from whom, and when, to maximize the buyer's overall satisfaction. Satisfaction, formalized as multiattribute utility, includes attitudes toward quality, reputation, and risk. The Prequote-Quote-Rescind (PQR) protocol communicates probabilistic and temporal information on the future prices and availabilities of items. A comparison set, defined as a set of bundles in which all items are available for a fixed interval and their prices are known, determines future intervals when purchase decisions will be fully informed. A decision procedure is provided that makes effective use of comparison sets and improves a buyer's expected utility compared with a naive decision procedure.