45 ms·
Is anyone working on making Wikipedia computable a la Wolfram Alpha?
A first step would be to have public domain source code associated with articles where it would make sense. Moreover, you would be able to execute this code via the web on Wikipedia data/examples possibly even with an animation explaining what is going on.
A second step would be to allow you to easily combine computations via the web, though not necessarily using a natural language interface.
UPDATE: The emphasis would be on the open source code, animations to understand it, and ways to combine this code to perform more complex computations in a practical and simple way (not a natural language interface). People could get the data from more trusted sources if necessary.
- phatbyte 17y agoThat will be awesome, except for one thing. Wikipedia is editable by everyone, so the data results couldn't be 100% accurate.
- cema 17y agoFew things are 100% accurate. More specifically for wikipedia, the results will not be 100% repeatable. Not necessarily a problem, but something to take into account.
- fleitz 17y agoPowerset aka. Bing
- paulgb 17y agoThere is an open source project extracting data, but I forget what it is called. Freebase is doing it commercially (they maintain a separate database, but I believe they have import data from wikipedia too). They provide an API and (IIRC) a data dump, but with a more restrictive license than Wikipedia itself has (again, IIRC).
- babyshake 17y agoFreebase is mostly focused on creating and curating a very clean programmatic version of Wikipedia, but it is definitely not computational in a way comparable to Wolfram Alpha.
- paulgb 17y agoHow so? I always assumed Wolfram Alpha was an NLP and data presentation layer on top of a Freebase-like database. Freebase doesn't have that extra layer, but structured data is surely part of the battle.