Is weighted tagging a better option?
I am trying to get a feedback from various communities before starting the program. There are many such examples where one object means more than one thing and tagging does a wonderful job. I still find a limitation. Let me take an example to explain.
I want to tag an auto loan website which offers some information about credit scores. Also it runs a forum, a community, blogs, wiki and few more sections like auto insurance. As a user I am not happy when I am tagging the website with “auto loan, auto insurance, credit score, forums, blogs, wiki, community”. I know that site is more about auto loan than about auto insurance or credit score, wiki is just a small part of it. In long term with few hundreds of tags (lets take 200) we may be able to see a clear difference. May be that
Site A
---------
95% (190 tag count) tagging the site with auto loan
40% (80 tag count) tagging the site with auto insurance
20% (40 tag count) with forums or wiki e.tc
Here I will face another problem that is if I have another site on auto insurance with lesser tag count (say 50).
Site B
---------
95% (47 tag count) for auto insurance
So which site is more recommended for auto insurance according to the tagging system? I need an algo to find out. I suggest a weighted tagging option.
Say an option of tagging with a weight from 1 to 5.
Site A
-------
95% with average 4 weight (190x4 = 760 tag count)
40% with average 1 weight (80x1 = 80 tag count)
Site B
-------
95% with average 4 weight (47x 4 = 188 tag count)
As the tag count grows the number converges to a better estimation. What do you say? Please let me know your suggestion and then I will start my work on it.
Thank you for your time.
Aji
|
Linear Mode
