Personalized Search
- 详细技术说明
- This technology is a personalized search toolbar that adapts to user interest based on search context and history, returning results that have a high probability of reflecting the user's original search intent without requiring any explicit user feedback.
- *Abstract
-
This technology is a personalized search toolbar that adapts to user interest based on search context and history, returning results that have a high probability of reflecting the user's original search intent without requiring any explicit user feedback. Based on research from the User-Centered Adaptive Information Retrieval (UCAIR) project in the Department of Computer Science, this technology complements search engines by providing more relevant results to the user.
DESCRIPTION/DETAILS
A major limitation of most existing retrieval models and systems is that they either ignore search history and context or require user-specific parameters to bound a single search session. The Personalized Search engine maximizes background/ implicit feedback from previous searches and clickthrough information to improve search accuracy with no extra effort required from the user.
For example, a computer programmer, a coffee connoisseur and a tourist may all enter the same word, "java," into a search engine, but each may be looking for entirely different information. The Personalized Search engine would return results for each based on previous searches and user interest, generating a contextually rich, optimized search.
The user-centered adaptive information retrieval (UCAIR) search engine takes into account long-term statistical trends and general interests and then pairs that with session-specific data based on recently-viewed documents and clickthrough trends. During browsing when a user clicks on the "Back" button of the web browser or the "Next" link of the search result page to view more results, the updated results displayed would be narrowed-down from the original.
How it works:
The Personalized Search engine technology works by inferring a user's interest from search context. It accomplishes this by running a proprietary algorithm that incorporates fixed coefficient interpolation, Bayesian Interpolation, on-line Bayesian Updating and Batch Bayesian Updating into one efficient and intelligent statistical model.
Essentially, UCAIR makes a note each time a user conducts a search, then it remembers what links the user has decided to follow. The UCAIR toolbar has three major components: 1. The implicit user modeling module captures a user's search context and history information, including the submitted queries and any clicked search results, which it uses to infer search session boundaries. 2. The query modification module selectively improves the query formulation according to the current user model. 3. The result re-ranking module updates any unseen search results immediately each time the user model changes.
APPLICATIONS
- Plug-in for web browsers: Popular add-ons to web browsers, such as the ubiquitous Google Toolbar, make internet searches a less troublesome task but require user input. The UCAIR search toolbar runs in Internet Explorer, silently filtering unwanted material and helping users quickly get to the information they need.
- Improve web-search results: Personalized Search generates highly accurate web searches based on user trends and recent activity, which refines its search results with each subsequent user action.
- Balanced decision making: The engine's return algorithm pairs long-term user trends and interests with session-specific query history and clickthroughs to generate a comprehensive picture of the desired results.
- Seamless searching: When engaged, the algorithm generates a background probabilistic model independent of the user, making high-probability web searches as accurate and as easy as possible.
- Client-side search agent: By storing search profiles and histories on the user's own computer, Personalized Search alleviates the privacy concerns that have surfaced surrounding search companies' attempts to develop similar methods.
For more information about this technology, please contact the University of Illinois at Urbana-Champaign Office of Technology Management at otm@illinois.edu.
- *IP Issue Date
- None
- *IP Type
- Utility
- 国家
- United States
- 申请号码
- 8095582
- 国家/地区
- 美国
