Modelling and Detecting Changes in User Satisfaction

The 23rd ACM Conference on Information and Knowledge Management (CIKM 2014) |

Informational needs behind queries, that people issue to search engines, are inherently sensitive to external factors such as breaking news, new models of devices, or seasonal changes as `black Friday’. Mostly these changes happen suddenly and it is natural to suppose that they may cause a shift in user satisfaction with presented old search results and push users to reformulate their queries. For instance, if users issued the query ‘CIKM conference‘ in 2013 they were satisfied with results referring to the page cikm2013.org and this page gets a majority of clicks. However, the conference site has been changed and the same query issued in 2014 should be linked to the different page cikm2014.fudan.edu.cn. If the link to the fresh page is not among the retrieved results then users will reformulate the query to find desired information.

In this paper, we examine how to detect changes in user satisfaction if some events affect user information goals but search results remained the same. We formulate a problem using concept drift detection techniques. The proposed method works in an unsupervised manner, we do not rely on any labelling. We report results of a large scale evaluation over real user interactions, that are collected by a commercial search engine within six months. The final datasets consist of more than sixty millions log entries. The results of our experiments demonstrate that by using our method we can accurately detect changes in user behavior. The detected drifts can be used to enhance query auto-completion, user satisfaction metrics, and recency ranking.