DMOZ is so weird

Its search is weak

[ebizcamp]

Although DMOZ has 5 million links (or more?), its search function is very weak. Say, search paypal, you even cannot get paypal's official site; search TOEFL, No Open Directory Project results found!

Am I right?

[OrlandoTodd]

Yea.. I think they've been having trouble with the search function for sometime.

If you do the search Google's Directory, it will search Google's copy of Open Directory. Paypal is in there under:
Computers/Software/Internet/Servers/Shopping/Auctions/Payments/

[ShootinBlanks]

Great tip!

Thanks, I've been pretty much ignoring them because of the lousy search, but with this tip they seem like a worthwhile deal. I wish they were a bit faster, though.

Cheers,

Ken :o)

[choster]

Only the rarest of users actually visits dmoz.org, and so the project's resources are generally directed internally toward category editing and maintenance and related features. The dmoz.org search is not expected to be a refined end user product, just like dmoz.org itself never had the goal of becoming the next It site (otherwise, why choose the pukey green? :) ). The expectation is that most will use the feature- and presentation-enhanced versions offered by data licensees such as Google.

[OrlandoTodd]

Exactly. Dmoz is only really used to check and see if your site is listed. You won't get traffic from Dmoz, but you will get better traffic from other search engines if you are listed there.

[SEOMike]

From their "suggest a URL" page...

"We aren't a search engine..."

So, there ya go.

[Macro]

From their "suggest a URL" page...
"We aren't a search engine..."

But they do offer a search box and an advanced search box. Shouldn't they just stop pretending? ;)

[pixel_juice]

>>Only the rarest of users actually visits dmoz.org

But some of the most valuable visitors do just that. Dmoz should fix their search and other inexplicable problems.

[Macro]

ebizcamp, the search function is not "weak". It's broken. It used to work like a normal search before but it doesn't anymore. Hasn't for weeks/months!

And that's not the only thing that's broken at DMOZ. They're still stuck in the 1990s. They don't have a basic automation system to notify webmasters when they sites have been accepted or rejected. Which means that thousands (millions?) of webmasters are using that search function everyday to see if their sites are listed, finding that the search doesn't work, and then drilling down to the category to see if their sites are there. What a waste of resources/bandwidth! But DMOZ won't "fix" that either. Why? Because they don't have the resources. If they've ceased caring for their users, don't expect them to bother about webmasters wanting listings.

And don't expect the search to be fixed anytime this week.

[pixel_juice]

>>Because they don't have the resources

Dmoz has a vast amount of resources at it's disposal. I can't see how this is the reason. The only reason I care is because Dmoz should be great.

[ebizcamp]

Macro,

Google has its API and people can develop own Google search application. I wonder if we can integrate Google API (search web directory instead of web pages) in our DMOZ sites. So, when users browse, he get DMOZ lists, when he search, he get Google Web Direcotry.

Google Web Directory is a few week(months?) lag behind of DMOZ. But basically, these two are identical.

[ebizcamp]

2. Can Google APIs be used to access Google Groups? Image search? Directory search?

No. The Google Web APIs service can only be used to search Google's main index of more than 4 billion web pages.

Just searched API Google FAQ. No way.

[blaze]

I think dmoz wants to ensure that it's not a competitor to any of the search engines. This is what allows (correctly) to keep it's mandate as a 3rd party resource for directory data.

[g1smd]

The search got broken at least partly as a consequence of converting all of the underlying directory data over, from various character encodings, to UTF-8 instead.

Over the last 6 months all of the category names and paths, and all of the site URLS, titles, and descriptions have been converted. Many were in ISO-8859-1 but not all converted cleanly (there were at least half a dozen different types of hyphen in use in the data, for example). Many other categories were in all sorts of other encodings, and a small number had no default encoding defined at all.

At the same time as converting all of that over to UTF-8, the character checking routines were improved to not allow any non-valid data into the database from now onwards.

Over half a million "errors" had to be fixed up either by running scripts, or by hand-editing the entries. The last 4 RDF files were down to just one or two errors in each one.

Once there are no errors in the RDF I expect that search will improve, but there is other work needed to be done to it, and I don't believe that is one of the highest priority jobs for the staff programmers.

Use Google's search function for the moment. The ODP RDF has been produced weekly for well over a year now, so downstream users now always have a recent RDF to update their site from.