Currently pursuing my Doctoral degree in Information Systems at Questrom School of Business, Boston University, advised by Prof. Nachiketa Sahoo and Prof. Chris Dellarocas. Previously, I received B.S. from Wuhan University in 2006 and Ph.D. of Computer Science from Institute of Computing Technology, Chinese Academy of Sciences in 2012.
595 commonwealth ave, office 519
Boston, MA 02215
The diversity of a set of recommendations can improve consumers’ satisfaction with a personalized recommender system. However, diversifying a list of items for a one-shot recommendation sacrifices relevance, which can reduce its value. We identify a popular scenario, sessions of online news consumption, where one can increase the diversity of recommendations over an entire session while improving the relevance of each recommendation within the session. Our approach is based on a multi-category utility model that captures consumers’ preference towards different types of content, how quickly they satiate with one type and substitute it with another, and how they trade off potentially higher value from their own costly search efforts with the convenience of selecting from a recommended list to find new content. Taken together, these three elements enable us to characterize how utility maximizing consumers construct diverse “baskets” of content over the course of each session, and how likely they are to click on content recommended to them.
We estimate this model using a clickstream dataset from a large international media outlet and apply it to determine the most relevant content at different stages of online sessions. We find that recommendations based on our approach are not only more diverse over a session, better matching the diversity sought by individual consumers, but also 6%–14% more accurate than recommendations by optimized alternatives. Using a policy simulation, we estimate that following our approach would cause visitors to read 57% additional articles at the studied website, which has direct revenue implication for publishers.
The idea of path-to-purchase is often discussed in Research and Practice. We develop a method to identify the most common paths-to-purchase using multi-variate time-series datasets collected from CRM systems. The proposed approach involves using a generalized Vector Auto-Regression model to capture consumers’ movement from one activity to another (steps of the path), and at the same time identifying groups of consumers who have similar paths.
The proposed approach is evaluated on a large customer touch point dataset made available by the Wharton Customer Analytics Initiative. The multi-time period paths show distinct shopping behavior of customers in different groups. The value of knowing such paths is illustrated through an example application of targeted catalog mailing.Go to Top