Forum Moderators: open
1. Five Open Directory Categories, and
2. A number of Open Directory Sites taken from the categories above
Question 1 is how does Dmoz decide which categories to display for any given search term?
Question 2 (probably directly related) is how does dmoz decide which sites should appear here?
When dmoz gets down to an actual category results are listed a-z - but before that happens it must be using some ranking algorithm to display suitable categories and suitable websites.
Any thoughts?
Results seem to be weighed with most (maybe 50%?) preference on the site title, and smaller percentages for site description, category names, and finally, there also seems to be something with "clustering". What I mean by this is that if most of the sites in a category have the keyword "foo", but several don't, those that don't may still appear high in the rankings simply because it's assumed that since they're in the same cat as the others, it'll have similar content.
I haven't explored it enough to come up with exact figures, but those seem to be the factors involved.
G.
The cats returned usually have one or all the search phrases in the cat title, or it has a matching site in the cat. The sites returned have at least one of the words included in title, cat, description or url.
I've read that the ranking doesn't have a ton to do with relavancy.
Good luck,
rmjvol
"The ODP maintains that their search is for editors only. They place no relevancy in their search function. "
Although they say this the fact remains that ODP search IS used by normal searchers - and the search facility is carried over to many/most of its 200 partner sites [dmoz.org...] - I suspect most traffic coming from ODP is by users using the Search facility.
I wonder if the categories are chosen first on the results page - and THEN the sites are selected from these categories. Or is it the other way round - Dmoz selects the most revelevent sites and then lists the cats that contain these sites?