The Noisy Channel

 

Wolfman vs. Googzilla

May 5th, 2009 · 5 Comments · Uncategorized

What’s not to love about a good fight? Check out David Talbot’s “Wolfram Alpha and Google Face Off” in Technology Review. I don’t come away with a sense that I’ll regularly use either Wolfram Alpha or Google Public Data, but it’s nice to start seeing people off-road with them and compare the results. The Wolfram Alpha launch is supposed to be this month, so presumably we’ll all be able to do that in a matter of weeks, if not days. Google Public Data is available already, intergrated into web search.

Sadly, neither of these guys seems interested in providing a non-NLP interface. In my view, that would be far more useful. But I suppose it’s not what sells papers.

5 responses so far ↓

  • 1 jeremy // May 5, 2009 at 8:11 pm

    What’s not non-NLP about the interface? From the Tech Review article, it seems like David Talbot was just entering noun phrases and noun conjunctions, e.g. of the form [microsoft apple]. In other words, a keyword mash.

    That is hardly NLP.

  • 2 Daniel Tunkelang // May 5, 2009 at 8:33 pm

    Let me rephrase–I want to see interfaces that allow for structured queries, and ideally that help users / applications discovers related structured queries. Instead, both Wolfram Alpha and Google seem determined to figure out the structure on their own.

    The functionality here, in my view, should be more like Excel and less like (the original) AskJeeves.

  • 3 Christopher // May 5, 2009 at 10:42 pm

    I seem to agree with you an awful lot. 🙂

    They should be concentrating on a Structured Natural Language Query, which is structured but within the structure understands synonyms so multiple predicates can be used (“How Many”, “Count” are equivalent).

    Oddly enough if implemented correctly a structured query language with “smart”, minimal natural language support can appear to be open ended. I still parse for verb usage, noun types but this is used in conjunction with a structured language that is representative of the semantic layer which is used in the exploratory interface, etc.

    Secondly the structure exposed in the query box should be reflected in an exploratory interface which is populated along with the results. This is an area where semantic web structures can really come in to play.

    Disclosure: SNLQ or Structured Natural Language Query is what I call the query portion on my interface & it follows the above methods so I am completely bias in my views.

    Though even with my bias I think after the amount of research & testing I have done of both pure NL & structured NL query capabilities I’m correct. 🙂

  • 4 jeremy // May 6, 2009 at 2:17 pm

    I want to see interfaces that allow for structured queries, and ideally that help users / applications discovers related structured queries. Instead, both Wolfram Alpha and Google seem determined to figure out the structure on their own.

    Ok, I get it now. And I totally agree as well.

    And from an HCIR perspective, it’s an absolute no-brainer. Why? Because having the system help you formulate these structured/semi-structured requests does two things: (1) It helps you form better queries, and thus find more relevant information, and (2) It allows you to understand that the system understands what it is that you’ve expressed.

    The second part of that cannot be overstated.

  • 5 Christopher Rines // May 6, 2009 at 10:16 pm

    Bang on Jeremy. Structure is key from an input, exploration point of view and from an explanation point of view.

Clicky Web Analytics