Ask YC: Ideas for FluidDB?

7 pointsby jdunckvor 17 Jahren6 comments

FluidDB is a centralized but scalable datastore. In it, the meaning of a given object is derived from the tags placed upon it-- and any user can place namespaced tags on any object. The idea is that ad-hoc standards for object IDs and tag names will emerge through use. I think this is a pretty novel shared-data concept and have had a couple ideas for it, but I'm sure news.yc will have additional novel applications. (I am a fan of FluidDB and its creator, but am otherwise unaffiliated.)

What interesting apps can you imagine with this shared-data system?

You can get a high-level idea of it here: http://abouttag.blogspot.com/2009/08/guardian-1000-novels-everyone-must-read.html

You can get into the details here: http://doc.fluidinfo.com/fluidDB/

6 comments

orlinvor 17 Jahren

I plan to build a visual FluidDB query editor with DSLs on top. One may call it "language workbench". Here is http://fluidmix.org/ the site. Will be doing the query tree interface next. I believe FluidDB is a great platform for search and I intend to take advantage of it.

terrycojonesvor 17 Jahren

Hi Orlin

I hadn't been to fluidmix since you first mentioned it. the site looks really nice.

jdunckvor 17 Jahren

So, the lack of response is a bit surprising. Am I wrong that FluidDB is novel? Or are people just having a hard time with the idea because it is novel?

terrycojonesvor 17 Jahren

Hi Jeremy

The information model takes some time to sink in - at least for most people (including me). Examples will help people to "get" it better. We have one coming out soon (next couple of weeks). I can give you an early look if you're ineterested.

terrycojonesvor 17 Jahren

I guess I'll get the ball rolling. We're building (and close to releasing) an app that's focused on Twitter users. It has FluidDB objects that each correspond to a Twitter user. Because the FluidDB objects aren't owned, other apps can put their tags onto the objects too. The query language lets you search on objects based on any tags (that you have read permission for), so we're hoping we'll soon see a wide variety of tags on these objects, which will give us a bunch of new ways to identify users on Twitter, to tag them arbitrarily, etc. The are obvious things like ratings, or people you've met, but there are also lists (a tagged set of users is just a list, more or less) including queries that cross lists, and things like http://tunkrank.com scores about Twitter users. FluidDB gives a place to put all that stuff - without requiring you to ask permission or for your needs to be anticipated - and it gives you a way to query on it too.

That's just a first app. We'll open source it. I can think of many more, as you might hope :-)

Terry Jones (disclaimer: I work for Fluidinfo, makers of FluidDB)

otoburbvor 17 Jahren

Mobile carriers store subscriber information, and this causes all sorts of problems with mobile device characteristics, roaming and other special attributes that only the carriers control.

It would be great if each subscriber could own their own profile and have it synched with their own mobile device. This would require carriers to inter-operate with FluidDB, but since many of them use LDAP, it wouldn't take much to have this type of interface made available via a FluidDB client library.

This is framework of a simple application that I've been kicking around.