Browse Prior Art Database

Tool for increasing quality of internet searching

IP.com Disclosure Number: IPCOM000149696D
Original Publication Date: 2007-Apr-05
Included in the Prior Art Database: 2007-Apr-05
Document File: 2 page(s) / 75K

Publishing Venue

IBM

Abstract

The invention disclosure refines the results in a internet search that a user makes. At high level, our solution tracks the time a user expends in any of the links of the results and based on that is able to rank the level of accuracy of any result for this search. Usually when a user enters a query in an internet searching engine he/she is returned with a list of links as results for the search parameters introduced. Then, the user (usually) clicks on the links on a sequential way (one by one) looking for the desired information. If the visited link has something to do with the query entered by the user, the user will expend some time at it, reading and looking at the site. But when the result has absolutely nothing to do with the query, the user expends no time at this site and returns to the results list in order to visit another link. The main idea of this invention is to have a way to measure the time spent between links (inside the same search) and rank each link based on the user behavior (i.e. time spent between different visits). Therefore, after several low ranking values for a specific link (associated with a specific query), this link is taken out from this query and the query itself becomes more accurate for future uses. In this case, the user experience is enhanced because increased accuracy of the results.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 1 of 2

Tool for increasing quality of internet searching

The idea is that if a user gets a list of links as a result of a search, usually he/she clicks on one link and if the web page coming has nothing to do with the search or the information is not good enough, he/she clicks in the next link in a short period of time. By measuring this time you can have a notion of the accuracy of the link to the search done.

The invention is implemented as a piece of software that can run:
1.- as a plug-in in a proxy (or transparent proxy) users go through; or
2.- embedded (or as a plug-in) at the user web browser.

In the first case (as plug-in in the proxy) it should work as follows:
A user goes to an internet search engine (i.e. www.google.com) and enters the query parameters he/she wants to search for. Through the proxy, the query arrives to the search engine that builds answering back to the user an HTML page with a list of URL's. Proxy plug-in should hook the answer and store the results page (with the query parameters entered) in order to correlate a future web page request from that user with one of the search results provided by the search engine.

Once the user looks at the results list he/she decides to visit one of the links clicking on it. This request comes to the proxy that recognize it as one of the results of a previous query, and make this request to the web server (if it has not stored the page from a previous visit). At the moment the proxy delivers the web page to the client (once the page has been returned by the web server), it stores the time as T0 for that result for that query for that specific user.

Then when the user clicks on another result link, the request arrives to the proxy that detects it's another result from the same query and calculate the time delay between T0 and this moment (T1). Usually this is the time the user spends in the previous link and is the measure of the interest of this link.

As each user visits different links, for each query proxy plug-in will have a set of measurements stored. As a particular case, it might be some links showed in the results page with higher priority* which are not visited by the user (although the user visits others). For those links we can assume that the time spent (by the user) is 0. When a search engine delivers a result page on a specific query, it gives priorities of each result, showing what it thinks is more suitable for the query in the first positions of the re...