November 10th, 2004


Structured Information Retrieval

I've just finished an exhaustive search for a good PDA and realized that there are many applications that could be easily done using already available technology and data but that either don't exist or that I am not aware of. The general theme here is criteria-based search. The fact that many people have wireless PDAs, web-enabled cell-phones, and ultra light laptops makes all of the below applications extremely relevant.


  • Restaurant search: given time of day, location and other optional parameters (cuisine, price range) find the restaurants that would be open and close. I run into this problem all the time - it's 10pm, but we want to go out, where's the nearest restaurant? Yahoo, CitySearch already do searches based on location and the other parameters, but not time-specific searching.

  • Movies: given time of day, location and optional parameters (genre, viewer rating) find all possible movies to go to right now, considering driving time to the theatre.

  • Route Navigation: enhanced route finder (improvement for GPS, MapQuest, etc.) that would allow the user to enter driving speed on highways and city streets (thus allowing the user to individually adjust the weight of these options), as well as taking into account time of day, so as to consider congestion (even at the most general level, without real time information about traffic jams) when picking a route.

  • Product Selection: I believe that any and all stores simply must have a product selection wizard, different from the more common "product expert." The "expert" tends to ask you "what type of user you are? - professional, student, traveler" and I never believe their results. The "wizard" will simply list all of the available options for a given type of product, allowing you to pick only the ones you care about and comes up with a list of matches, including incomplete matches. DPReview's feature search for digital cameras is exactly what I mean, but I couldn't find a similar tool for other types of products (e.g., PDAs). How come?!? Is there a huge market for this out there?

  • Recipies: given a list of ingridients and optional parameters (cooking time, complexity rating, number of calories), suggest a list of dishes. Same could be done for drinks. Here, too, a simple keyword-based search in a database of recipies would not work.

These are just some examples, I am sure it's to come up with many more. The point is that the data (the hard part) is already out there, it's just not utilized properly. Of course, it might just be me being ignorant about existing technology.