A Personalized Search Using User’s- Profile
Naeem Naik1, L.M R.J. Lobo2
1Mr. Naeem Naik, Walchand Institute of Technology, Solapur, India.
2Prof. L.M R.J. Lobo, Walchand Institute of Technology, Solapur, India.
Manuscript received on September 22, 2013. | Revised Manuscript received on October 10, 2013. | Manuscript published on October 30, 2013. | PP: 55-57 | Volume-3, Issue-1, October 2013. | Retrieval Number: A2148103113/2013©BEIESP
Open Access | Ethics and Policies | Cite
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Users interest is an important area in the field of IR that attempts to adapt ranking algorithms so that the results returned are tuned towards the searcher’s interests. In this work we use user data to build personalized ranking models in which user profiles are constructed based on the user’s tagging data over a topic space. However, instead of employing a human-generated ontology, we use novel latent topic models to determine these topics. This means that the topic space used is determined based purely on the tagging data itself and therefore does not require human involvement to determine the topics. Our experiments show that by introducing user profiles as part of the ranking algorithm, rather than by re-ranking an existing list, we can provide personalized ranked lists of documents which improve significantly over a non-personalized baseline. Further examination shows that the performance of the personalized system is particularly good in cases where prior knowledge of the search query is limited. This is especially useful as these are the cases where we are unable to determine when same tag has totally different intention.
Keywords: Image search, Metadata, optimization.