Author |
Topic: Feedback Sought on CollaborativeRank Search Engine (Read 258 times) |
|
amichail
Senior Riddler
Posts: 450
|
 |
Feedback Sought on CollaborativeRank Search Engine
« on: Mar 24th, 2005, 9:41pm » |
 Quote  Modify
|
There is now a search engine available to experiment with the idea of having people collaboratively determine the rankings to various web search queries.
Please feel free to play with it:
http://www.cse.unsw.edu.au/~amichail/collabrank/
It currently includes the computers, science, and games sections from the open directory. You can of course add your own URLs on other topics.
There is even a high score list that shows you which users are most helpful in their ranking suggestions.
The system uses something I call "CollaborativeRank", which is a circular PageRank-like measure assigning importance to both people making ranking suggestions and the (query, url) pairs being scored.
In fact, the high score list is simply a ranking of people by their CollaborativeRank.
I plan to compute CollaborativeRank daily.
I am particularly interested in ways to make this search engine more usable.
One obvious problem is that a site you are searching for might rank very low, but the UI does not give you an easy way to find it so that you can "suggest higher rank".
At the same time, allowing people to "suggest higher rank" without even looking at the ranking for a query may not be a good idea as there might be other better sites for that query.
|
| « Last Edit: Mar 24th, 2005, 9:43pm by amichail » |
 IP Logged |
DropZap - a new kind of block elimination game
|
|
|
amichail
Senior Riddler
Posts: 450
|
|
Re: Feedback Sought on CollaborativeRank Search En
« Reply #1 on: Mar 26th, 2005, 4:05am » |
Quote Modify
|
In case anyone is interested, here is a more complete description.
==
CollaborativeRank is a search engine that allows its users to make ranking suggestions and rewards them whenever those suggestions are subsequently liked by other users.
If you tell the system to rank a URL higher with respect to a query and subsequently others make the same suggestion, then you will be rewarded by taking your ranking suggestions more seriously.
This is done by computing "CollaborativeRank", which is a circular PageRank-like measure that assigns importance to both people making ranking suggestions and (query, url) pairs.
Usage is simple. Try a query and if you don't like the results, suggest a URL that should rank highly for that query.
For example, say you search for "wysiwyg editor", but don't like the ranking of search results. Moreover, suppose you know about TeXmacs, which is an excellent wysiwyg scientific text editor.
In that case, you could use "wysiwyg editor" as the main search query and "texmacs" as the secondary query. You will then be shown where TeXmacs comes up in the ranking for the "wysiwyg editor" query.
If you feel it is ranked too low, simply tell the system to rank it higher by clicking on the up arrow for the TeXmacs URL. (You will need to register to make ranking suggestions.)
Finally, note that you can easily browse your higher rank suggestions, which means that you can use the system to keep track of your favorite URLs, much like any bookmark manager -- but here, your list of favorite URLs can have an impact on everyone's search rankings.
|
| « Last Edit: Mar 26th, 2005, 4:56am by amichail » |
IP Logged |
DropZap - a new kind of block elimination game
|
|
|
|
RIDDLES SITE
WRITE MATH!
Home 
Help 
Search 
Members 
Login 
Register
Reply
Notify of replies
Send Topic
Print