Browsing by Author "Sarkar, S."

Now showing 1 - 4 of 4

Metadata only
Accelerated learning of user profiles
(Informs, 2011-02) Atahan, Pelin; Sarkar, S.; Sectoral Education and Professional Development; DEMİRCİLER, Pelin Atahan
Websites typically provide several links on each page visited by a user. Whereas some of these links help users easily navigate the site, others are typically used to provide targeted recommendations based on the available user profile. When the user profile is not available (or is inadequate), the site cannot effectively target products, promotions, and advertisements. In those situations, the site can learn the profile of a user as the user traverses the site. Naturally, the faster the site can learn a user's profile, the sooner the site can benefit from personalization. We develop a technique that sites can use to learn the profile as quickly as possible. The technique identifies links for sites to make available that will lead to a more informative profile when the user chooses one of the offered links. Experiments conducted using our approach demonstrate that it enables learning the profiles markedly better after very few user interactions as compared to benchmark approaches. The approach effectively learns multiple attributes simultaneously, can learn well classes that have highly skewed priors, and remains quite effective even when the distribution of link profiles at a site is relatively homogeneous. The approach works particularly well when a user's traversal is influenced by the most recently visited pages on a site. Finally, we show that the approach is robust to noise in the estimates for the probability parameters needed for its implementation.
Metadata only
Composing offer sets to maximize expected payoffs
(Digital Commons, 2016) Atahan, Pelin; Johar, M.; Sarkar, S.; Sectoral Education and Professional Development; DEMİRCİLER, Pelin Atahan
Firms are increasingly using clickstream and transactional data to tailor product offerings to visitors at their site. Ecommerce websites have the opportunity, at each interaction, to offer multiple items (referred to as an offer set) that might be of interest to a visitor. We consider a scenario where a firm is interested in maximizing the expected payoff when composing an offer set. We develop a methodology that considers possible future offer sets based on the current choices of the user and identifies an offer set that will maximize expected payoffs for an entire session. Our framework considers both the items viewed and purchased by a visitor and models the probability of an item being viewed and purchased separately when calculating expected payoffs. The possibility of a user backtracking and viewing a previously offered item is also explicitly modelled. We show that identifying the optimal offer set is a difficult problem when the number of candidate items is large and the offer set consists of several items even for short time horizons. We develop an efficient heuristic for the one period look-ahead case and show that even by considering such a short horizon the approach is much superior to alternative benchmark approaches. Proposed methodology demonstrates how the appropriate use of information technologies can help e-commerce sites improve their profitability.
Open Access
Optimizing offer sets based on user profiles
(Social Science Research Network, 2009) Atahan, Pelin; Johar, M.; Sarkar, S.; Sectoral Education and Professional Development; DEMİRCİLER, Pelin Atahan
Personalization and recommendation systems are being increasingly utilized by ecommerce firms to provide personalized product offerings to visitors at the firms’ web sites. These systems often recommend, at each interaction, multiple items (referred to as an offer set) that might be of interest to a visitor. When making recommendations firms typically attempt to maximize their expected payoffs from the offer set. This paper examines how a firm can maximize its expected payoffs by leverag ing th e kn owledge of the profiles of visitors to their site. We provide a methodology that accounts for the interactions among items in an offer set in order to determine the expected payoff. Identifying the optimal offer set is a difficult problem when the number of candidate items to rec ommend is large. We develop an efficient heuristic for this problem, and show that it performs well for both small and large problem instances.
Metadata only
Sponsored search: the sum is larger than its parts
(2011) Ghoshal, A.; Sarkar, S.; Menon, S.; Atahan, Pelin; Sectoral Education and Professional Development; DEMİRCİLER, Pelin Atahan