Enhancing web revisitation by contextual keywords

Web revisitation is a common behavior supported by many web history tools. Taking advantages of access context (like time, location, concurrent activity), context-based search of previously accessed web pages is also being investigated, due to the fact that context under which information is accessed tends to be more easily to remember than content. To mimic users' memory recall, we present a way to automatically capture user's access context from user's concurrent computer programs, and manage it in a probabilistic context tree for each accessed web page in a life cycle. An algorithm for contextual keyword search of accessed web pages, together with a revisitation feedback mechanism, are also given. We evaluate the proposed method on synthetic data and through a 6-week user study. The comparisons of revisit precision and recall show our method outperforms the existing contextual search method YouPivot. In the user study, our method can also work as effectively as popular methods (like bookmark, browse history) in recall rate (over 90%), while with less average time cost (16.25 seconds) than that (38.66 seconds) of those methods to complete a web revisitation task.