Proactive search: Using outcome-based dynamic nearest-neighbor recommendation algorithms to improve search engine efficacy

The explosion of readily available electronic information has changed the focus of data processing from data generation to data discovery. The prevalent use of search engines has generated extensive research into improving the speed and accuracy of searches. The goal of this research is to accurately predict user behavior as a means to proactively improve speed, accuracy, and predictability of search engines. The proactive approach eliminates query entry time and hence reduces the overall processing time, improving speed. Assuming success, the user locates an electronic resource of interest, improving accuracy.;Algorithms that have been shown to predict many vastly different aspects of user behavior exist in literature. Two common approaches are used in such prediction: statistical techniques and collaborative actions. This research extends the scope of proactive search by using search histories of users in building a predictive model. The proposed approach was compared to statistical and collaborative behavior models. The test results verified that search engine prediction is a viable approach and supports the intuitive notion that prediction is more successful when user behavior exhibits less entropy; i.e., when the actions of a user follow a consistent and recognizable pattern.;The benefits of the proposed approach go beyond improvement in performance and accuracy. As a result of working with search histories as sequences of resources, it is possible to predict a series of resources that a user will likely select in the immediate future. This makes it possible for search engines to return resource sequences instead of simple resources. Working with sequences allows the search engine user to more effectively locate information of interest. In the end, a proactive search engine improves speed and accuracy through prediction and sequencing of electronic resources.