Location

260-092, Owen G. Glenn Building

Start Date

12-15-2014

Description

Recommender systems aim to support users in identifying the most relevant items. However, there are concerns that recommenders may imprison users in a “filter bubble” by recommending items predominantly known to them. On the other hand, providing unconventional items may increase risks of not meeting users’ taste. Given this trade-off, we analyze the effects of consumers’ perceived levels of recommendation novelty and serendipity on perceived preference fit and enjoyment. We find that merely increasing the level of novel recommendations is disadvantageous. Instead, recommenders should provide more serendipitous recommendations as this leads to higher perceived preference fit and enjoyment. In addition, market and recommender technology characteristics must be taken into account, since they partially determine the level of novel and serendipitous recommendations. Our findings have significant implications for research as they add additional insights on users’ evaluations of recommender systems. For practice, our results support online retailers in developing better recommenders.

Share

COinS
 
Dec 15th, 12:00 AM

Escaping from the Filter Bubble? The Effects of Novelty and Serendipity on Users’ Evaluations of Online Recommendations

260-092, Owen G. Glenn Building

Recommender systems aim to support users in identifying the most relevant items. However, there are concerns that recommenders may imprison users in a “filter bubble” by recommending items predominantly known to them. On the other hand, providing unconventional items may increase risks of not meeting users’ taste. Given this trade-off, we analyze the effects of consumers’ perceived levels of recommendation novelty and serendipity on perceived preference fit and enjoyment. We find that merely increasing the level of novel recommendations is disadvantageous. Instead, recommenders should provide more serendipitous recommendations as this leads to higher perceived preference fit and enjoyment. In addition, market and recommender technology characteristics must be taken into account, since they partially determine the level of novel and serendipitous recommendations. Our findings have significant implications for research as they add additional insights on users’ evaluations of recommender systems. For practice, our results support online retailers in developing better recommenders.