Personalized Web Search

Personalized Web Search is an emerging concept in Web Mining. It is a method to improve search quality by customizing search results for users with individual information goals as illustrated in Figure 1. The objective is to provide users with accurate information they want. Many personalized web search algorithms are used for analyzing the user interest and produce the effective outcome. Examples of such algorithms are User profiling, Hyperlink analysis, Content analysis and Collaborative web search.

Fig 1 Personalization of Web Search using Web Mining (Source Ref. 1)

Generic web search engines follow a model of one size fits all. Generic web search returns same result to all users for identical queries regardless of their interests and needs. Moreover sometimes the queries for search entered by users are short and ambiguous. Like for example one user enters a query for 'mouse' to get details about the animal whereas another user enters the exact same keyword to get details about computer hardware device. This is where personalized web search comes handy, it organizes results differently for different users based on one for the following methods:-

1. User profiling - It is a process of identifying data about users' needs. A user profile maintains information about user needs and interests. It contains information like age, gender, language, education, address, country, interests, search history, most visited pages and favorite web sites. Personalized web search can be implemented both on client side and server side.
2. Content Analysis - It is a method of checking content similarity between web pages and user profiles.
3. Hyperlink Analysis - It is a method of ranking the documents based on the linkage structure of the web. Various algorithms are used to calculate the importance of documents.
4. Community based/Collaborative web search - It is a method where search histories of users who have same interests/needs to the other similar users are used to filter or re-rank search results.

Personalization in web search has several challenges too, like for example users' privacy may get compromised in this method; also users' interests are not always static so users' profiling needs to be dynamic etc. Researchers are working to minimize these disadvantages. But there is no denying the fact that generic web search is steadily making way for personalized web search in the interest of high performance, better and faster results.

By: Ms. Shivani Gautam - Assistant Professor (CSE), Chitkara University, H.P.

References

1. Joshi, Chanchala, Teena Jaiswal, and Himanshu Gaur. "An overview study of personalized web search." International Journal of Scientific and Research Publications 3.1 (2013).
2. Kashid, Sheetal Vishwas, and Vishwakarma Pinki. "Personalized Web Search in Web Mining." IJAIS Proceedings on International Conference and Workshop on Communication, Computing and Virtualization ICWCCV. Vol. 1. 2015.