We've just started thinking about a new product (to be used internally at a company) that:
- Takes in multiple inputs (hundreds)
- Has a large set of templates (thousands)
- Has a large set of previously generated output (tens of thousands)
Generates new stuff.
- Location: Los Angeles. Band: Green Day. Nearest Venue: Staples Center. When: tomorrow Lowest Price: $49
- Template: "Hi there - we've got [Green Day] tickets available for [tomorrow] at [Staples Center] for [$49]!"
- Previous output 1: "Get tickets for [John Mayer]! [Saturday] at [Hollywood Bowl]. Only [$95]."
- Previous output 2: "Want to see the [John Mayer Show] at [Hollywood Bowl]. Buy now for [$95]!"
What we'd like the system to generate sensible output that can be used.
What we're focused on here is the UX challenge (not the technology or architecture or semantic language processing stuff).
We were thinking, for example, of some really visual ways of doing this, where you have a big set of targets on the screen, a big set of templates, and a big set of historical output. Some way of creating associations between them (pointing and clicking, dragging, etc.).
Then, some kind of simulation or preview, showing what the combination of things results in.
We also wondered about various visualization techniques, for instance, rending a set (inputs, templates, outputs) as some kind of physical object (like a cube or something) to maintain the relationship / binding of the items, if that makes sense.
Anyhow - we're not quite sure what direction to go. Any screenshots / examples of ways that similar problems have been solved would be appreciated.