The diverse back grounds of Web users lead to the notion of personalized search engines. Traditional search engines normally provide the same ranked search results to everyone. The concept of “one ranking fits all” of search engines is not effective, especially when the search is from a personal mobile device.
This personalization can be achieved by a machine-learning algorithm, which learns from implicit feedback provided by the users in the form of clicks. This research provides personalized rankings based on this collected implicit feedback data by using support vector machines (SVM).
The support vector machine builds a unique model according to the user-feedback data. Search results are ranked according to the unique model to meet each users specific needs. The proposed system greatly improves the overall ranking quality of search results
Source: Midwest Instruction and Computing Symposium
Author: Malvika Pimple | Wen-Chen Hu | Naima Kaabouch | Hung-Jen Yang