UFeedRevs - PowerPoint PPT Presentation

1 / 17
About This Presentation
Title:

UFeedRevs

Description:

Claim to index more than 19 billion pages. Today focus shifts from size to ... MySQL 4.3.x. NuSOAP 0.6.1. feedee (Screenshots) Interpretations from BETA tests ... – PowerPoint PPT presentation

Number of Views:36
Avg rating:3.0/5.0
Slides: 18
Provided by: akshay9
Category:
Tags: ufeedrevs | mysql

less

Transcript and Presenter's Notes

Title: UFeedRevs


1
UFeedRevs
  • Redefine Web Searches gtgt

2
Overview
  • Introduction
  • Related Work
  • UFeedRevs
  • System Flow
  • feedee
  • Interpretations of BETA tests
  • Future Research
  • Areas of Application
  • Questions or Feedbacks
  • Thanks

3
Introduction
  • Search Engines
  • Primary gateways of Information Access
  • Claim to index more than 19 billion pages
  • Today focus shifts from size to relevancy !

4
Related Work
  • Web click-through data
  • Merits Effortless, Shows Improvements
  • Demerits Inconclusive
  • Active Feedback of Effective Web Search
  • Demerits Increased User Involvement
  • Adaptive Web Search based on User Profile
  • Merits Improves Performance, Popular
  • Demerits Exhaustive, User Specific

5
UFeedRevs
  • User Feeds Relevant Results
  • Empower the User
  • Focus on relevancy and usefulness
  • Power of Collaborative Filtering
  • Re-define Web Searches

6
System Flow
  • Collect User Feedbacks
  • Filter User Feedbacks
  • Compute Feedback Rating

7
System Flow (contd.)
  • Compute Feedback Rating
  • Large Samples ( gt100)
  • Mode Evaluation
  • Statistical Dispersion around Mode
  • a - cutoff

8
System Flow (contd.)
  • Aging and Flushing
  • Value existing feedbacks
  • Consider new feedbacks
  • Flush the old ones

9
feedee
  • feedee is a pseudo-search engine.
  • More you feed it the better (fatter -) it gets
  • http//feedee.dbit.in

10
feedee (System Architecture)
11
feedee (Powered by)
  • Google Search API
  • PHP 4.1
  • MySQL 4.3.x
  • NuSOAP 0.6.1

12
feedee (Screenshots)
13
Interpretations from BETA tests
  • Suitable for theme-based searching
  • Overall increase in relevancy of results
  • Limited resistance to User Bias

14
Future research
  • More extensive testing in terms of
  • scale
  • audience
  • query categories
  • Recognizing areas in which this scheme can prove
    to be useful
  • Optimizing feedee

15
Areas of Application
  • Theme-based Search Engine
  • Local Transparent Layer
  • Query Knowledge Discovery
  • Potential for Non-text based searches
  • Document Search Engine
  • ...and many more

16
Questions or Feedbacks
  • Our quest to answer a few key questions ended up
    in more questions

17
Before we say Thnks!
  • Group Members
  • Akshay Surve
  • Manav Shah
  • Project Guide
  • Prof. Amiya Tripathy
  • Special Thanks
  • Dr. S. Krishnamoorthy (Director RnD, DBIT)
  • Dr. Revathy Sundararajan
  • Prof. D.D. Patil
  • Everyone who participated in BETA testing
    (feeding)
Write a Comment
User Comments (0)
About PowerShow.com