Chris Pollett > Students >
Vijaya

    ( Print View )

    [Bio]

    [Project Blog]

    [CS297 Proposal]

    [Deliverable1]

    [Deliverable2]

    [Deliverable3]

    [Document[Vijaya.doc]]

    [YioopCodeStudy]

    [Deliverable4]

    [CS 297 Report- PDF]

    [CS298 Proposal]

    [CS 298 Project Report- PDF]

    [CS 298 Final Presentation- PDF]

                          

























CS298 Proposal

Smart Search: A Firefox Add-On to Compute a Web Traffic Ranking

Vijaya Pamidi (vijaya_pamidi@yahoo.com)

Advisor: Dr. Chris Pollett

Committee Members: Dr.T.Y.Lin and Dr.Robert K. Chun

Abstract:

The main goal of this project is to improve the Web traffic ranking for the search Yioop search engine and giving user centric results. Search engines results are given to the user based on the ranking and indexing strategies followed by the search algorithm of a search engine. Currently, there are some tools available like www.alexa.com, www.ranking.com, www.compete.com which give analytic data for ranking web sites based on web traffic, the number of users who visit a site. Alexa provides the traffic rank for a web site based on two factors: The number of users that view a web site and the number of pages viewed. The main goal of our project is to create a Smart Search Firefox add-on for the Yioop search engine, an open source search engine developed by my project advisor, Dr. Chris Pollett. This add-on will provide similar analytic data to the Yioop search engine. Smart search considers three main factors to get a rank for a web site, these factors are: Number of users that view a web site. Probability that user leaves some other web site and enters the current one and the probability that user clicks on the link provided on the current page. Getting the user review for a result: if the user finds a particular result useful clicks on the + sign. A + will increment the rank of a page or web site Smart search takes these three basic factors into consideration and sends back data to the Yioop search engine periodically. With the results received from the Smart Search tool, the Yioop search engine refines the search results. Eventually, users would benefit from these better search results.

CS297 Results

  • Google's Page Rank Algorithm modeled for a 10X10 matrix.
  • Made a simple Firefox extension with alert functionality added.
  • Captured user activity by the XUL tool bar button.
  • Communicated with Yioop by sending the captured links to Yioop.

Proposed Schedule

Week 1: Jan26_Feb 1Getting CS298 Proposal approved and upload.
Week 2: Feb 2_Feb 8Study Firefox's SQLite and database design for user end captured data.
Week 3-4: Feb 09_Feb 22Capture user activity all three features, make configuration panel for the tool bar.
Week 5: Feb 23_march 01 Send the captured data periodically to Yioop
Week 6-7: March 02_March 08Re rank the Yioop index based on the user results received.
Week 8: March 09_March 15Send the results back based when a search query is given to Yioop.
Week 9: March 16_March 22 Conduct a result study and do the statistical analysis.
Week 10: March 23_March 29 Prepare a report of the results.
Week 11: March 30_April 05 Work on CS298 Report
Week 12: April 06_April 12Submit CS298 Report First Draft.
Week 13: April 13_April 19 Complete the report for committee review.
Week 14-15: April 20_ May 03 Final Report Submission and presentation slides.
Week 16: May 04_May 10 Present to Committee.

Key Deliverables:

  • Software
    • Deliverable 1: A Firefox component that uses an SQLite database to store the user end captured search history.
    • Deliverable 2: A component that sends data from Firefox to Yioop periodically.
    • Deliverable 3: A Yioop component that refines the ranking algorithm based on the received data.
    • Deliverable 4: A component that gives refined results to users for their search queries.
  • Report
    • Deliverable 5: CS298 Project Report
    • Deliverable 6: Project code Documentation.

Innovations and Challenges

  • Building a Firefox traffic toolbar that communicates with the Yioop open source search engine is innovative as the existing traffic toolbars communicate with closed source search engines.
  • Many downloaded instances of our toolbar can send results to the Yioop server simultaneously, so making the Yioop server scalable to handle these results is challenging.
  • Indexing these results on Yioop is challenging.

References:

[LM2006] Amy N. Langville and Carl D. Meyer. Google's PageRank and Beyond. Princton University Press. 2006.

[GBM2009] Gavin Bell. Building Social Web Applications. O'Reilly Media. 2009.

[KF 2007] Kenneth C. Feldt.Programming Firefox: Building Rich Internet Applications with XUL.O'Reilly. 2007

[SH 1998] Steven Holzner.JavaScript- Complete- 1998.

"http://developer.mozilla.org/en/docs/Building_an_Extension":Official page of Mozilla.