Browse Prior Art Database

Method, system, and apparatus for automatic keyword recommendation based on individual user behavior Disclosure Number: IPCOM000174084D
Original Publication Date: 2008-Aug-26
Included in the Prior Art Database: 2008-Aug-26
Document File: 7 page(s) / 81K

Publishing Venue



The present invention provides a computer implemented method of refining search result by keyword recommendation to filter unrelated items in search result, and the recommendation is based on both user’s input and analysis result of documents or results clicked by user in subsequent actions. The method includes four steps: retrieving initial search result based on user’s input; record and analysis the documents or results which are clicked by user; recommend keywords according to the analysis result and user’s original input; refine search result if one of recommendations is followed by user. The present invention has two main advantages in improving user’s experience. Firstly, it can be more efficient in understanding user’s search purpose because the recommendation is customized for each user’s separate search activity, and then return more accurate and refined search result. Secondly, the recommendation will help more users to be experienced searcher. The following characters make the present invention distinguish from other invention: 1. Recommendation is based on both terms user specified and analysis result of documents clicked by user in subsequent actions. So if the initial inputs are same, but following actions are different, then the keywords recommended may be diverse. 2. Recommendation keywords are generated dynamically. This character makes the present invention differ from those provide same recommendation for all users and users’ each search activity. 3. Recommendation keywords may take out some words from user’s initial input, or add some new words. In addition, user’s initial keywords may be wholly replaced.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 40% of the total text.

Page 1 of 7

Method, system, and apparatus for automatic keywordrecommendation based on individual user behavior


Fig. 1-a, 1-b and 1-c illustrates how the method interacts with users via a search sample.

Fig. 2 describes how the presented invention works.

Fig. 3 depicts a software architecture in which methods and systems may be implemented according to the present invention for user selection driven search query optimization.


101 102

Apple price


Apple - Education - Price Lists

K-12 and Higher Education Apple Professional Development, Curriculum Development, Product Training, and Technical Training Price List, 05/13/08 ...

3G iPhone catalyst behind Apple price target hike to US

Follow the markets in real time with FP Trading Desk, your up- to-the-minute look at Bay Street and beyond from Financial Post.

Shenzhen: apple price increased dramatically in Nanchang wholesale

                                              In the Shenzhen Nanchang wholesale market, a fruit retailer has mentioned that the apple price has increased by 50% this year due to the low production ...

iPhone revenue sharing pushing analysts to rethink Apple price targets

2007/11/9 ... One item that's been giving analysts trouble is how to predict the stock price from Apple now that it is getting revenue over time from the ...

Green Apple Fruit - Compare Prices, Reviews and Buy at NexTag ...

Green Apple Fruit - 243 results like the Kappus Soaps - Green Apple - Fruit Soap (oval) 4.2 oz, Scitec ISO Fruit Delite Green Apple,2 lb, ...

Next Page




Fig. 1-a


Page 2 of 7

In Fig.1-a, the initial search action begin from 101. User inputs the keywords in input box 101, then click search button 102, and search results will be shown by pages. For the sample shown in Fig. 1-a. User is interested inthe apple price, then he input keywords "Apple price", but the results contains several kinds of "Apple" price. In fact, he/she

which is related to his/her needs.

In Fig. 1-b, a recommendation message 105 is shown just after user's first two clicks. In this message, keywords "Apple price fruit" is recommended to user and this recommendation is based on the analysis result of search items 103 and 104

system will automatically analysis it together

                                                    After the analysis on documents 103 and 104, the system finds out that the two documents have another keyword "fruit" in common besides "Apple price", so new keywords "Apple price fruit" is recommended. User can follow the recommendation or ignore it. If user chooses the recommendation, then he/she clicks recommendation 106, and Fig. 1-c will be shown.

In Fig. 1-c, input box 107 is replaced with keywords "Apple price fruit", and the search results returned 108 are all fruit "Apple" price related which is user really wants, and then time can be saved from finding needed information page by page....