Someone code this please...

[icon]

I have a friend who was complaining about having problems finding information about graduate research programs in his field using Google and University websites in general. In response to these comments I started ranting and the rant turned into a sort of odd idea to create a Node Based interface to Google/Other Search Engines.

In my mind this would be a non-destructive, visual GUI to aid in performing, ordering and analysing multiple searches online. It'd be similar in look/feel to something like Shake [screenshot] or Maya's Hypershade Editor, with the focus being on utility nodes that connect to each other and eventually to a 'Print Node' which outputs your results in an AJAX interface that updates as you edit your nodes.

An example of how this would work follows, tailored to my friends problem:

Take the results of a Google search for "University Graduate Programs", turn it into a little visual [Search Node] in a Fash/JS/AJAX inferface,

Create another [Search Node] for the results of "Research in MyFieldOfInterest" on Google,

Create a new [Cross-Section Node] and drag the output of your previous nodes into the Cross-Section node's input[1] and input[2],

Create a new [Iterative-Search Node] for "Universities in [$ListOfCountries]" and fill in the array $ListOfCountries with the places you want to visit,

Create a [Prioritise Node] and drag output of previous Cross-Section Node into it's Input[1]. Drag the output of your Iterative Search node to the Priority[1] input on this node,

Create a new [Print Node] and plug the output of the Prioritise Node and watch the AJAX table fill up with your results.

The obvious advantage of a system like this is that it's non-destructive so you can change variables and view the new search results in real time. This allows for a modular tweaking/refining of your search.

I think there's also the room to create custom Social Networking / Social Media nodes and var-lists that people could contribute too. So for example "Counties Amazon Delivers Too" might be a contributed node/var. Or there might be ... "Universities Offering Graduate Programs in X", and if there isn't then the output of your search could be saved and shared as just such a node.

Aaaaanyway ... I thought it was a cool enough idea to handball on and see if anyone wants to pick it up. If you make bazillions of dollars from it you can buy me a case of Coopers

[heist]

i don't get it - what's this difficult new interface designed to achieve?

[icon]

I like the node/DAG model because it allows you to perform complex operations in a simple and intuitive way. It really shines when you've got a workflow-type set of operations and want to be able to change things and see the results in real time.

It's also easy to share those sorts of networks and develop your own methods for achieving results. For example you can share a Shake network for compositing but you can't share Photoshop layers (not as easily anyway) , instead you have to teach the PS technique.

There's lots of things I'd like to be able to do with my web searches (recording, comparing and piping the results being a few that pop to mind) but usually can't. I think this would be a simple way of doing those otherwise complicated things.

Although you might be right ... it could just be over-complicated and unnecessary. I do that a bit

[heist]

Originally Posted by icon

I like the node/DAG model because it allows you to perform complex operations in a simple and intuitive way. It really shines when you've got a workflow-type set of operations and want to be able to change things and see the results in real time.

It's also easy to share those sorts of networks and develop your own methods for achieving results. For example you can share a Shake network for compositing but you can't share Photoshop layers (not as easily anyway) , instead you have to teach the PS technique.

There's lots of things I'd like to be able to do with my web searches (recording, comparing and piping the results being a few that pop to mind) but usually can't. I think this would be a simple way of doing those otherwise complicated things.

Although you might be right ... it could just be over-complicated and unnecessary. I do that a bit

sorry - this is probably very obvious to you as you like to work this way, but i really only use web searches to find info. search -> click -> read and i'm done.

i'm not sure why someone would want to graph/tree/composite/record the results of a web search. can you give me a real world example?

is it like, so you can save a neatly tailored web search and see when new info becomes available in that specific area... like a feed?

[jexley]

Mate, that IS a bazillion dollar idea. Somebody's surely thought of something similar somewhere too (probably), so the question is just what you asked: Who's building it?

Me, I'd love to, except the word "node" makes me think of something a Hobbit would grow in his garden and wouldn't taste very good unless you add a lot of salt. Plus, I'm busy trying to build my own bazillion dollar ideas.

Good luck though, I like it and think its neat.

[tuna]

I have seen this done before, I think some french dudes have already done it.