4 ms·
Wolfram Alpha added to Bing
- theli0nheart 17y agoI wonder who's profiting more from this deal, Wolfram Alpha or Bing?
- noodle 17y agoit seems like a good idea for both of them. but in absolute terms, bing probably has more to gain.
- dandelany 17y agoWolfram Alpha has a lot more to lose... Microsoft is a gigantic company. This has gotta be Wolfram's #1 revenue source now, and they needed it badly.
- psyklic 17y agoAlpha is their #1 revenue source? Whatever happened to Mathematica?
- pchristensen 17y agoI assume you mean it's Wolfram Alpha's #1 revenue stream, not Wolfram as a whole.
- dandelany 17y agoI do.
- elblanco 17y agoAlpha is really struggling to find a successful use-case for its technology. It has a ridiculously powerful math engine behind a semi-working ontology (not saying that theirs is bad, just that all general ontologies are "semi-working") that's driving a rather poor natural language parser (again, pretty much all NLP's are "poor"). In effect they are trying to use the ontology as a poor AI to "understand" the NL. People who want to use Alpha have to deal with it through the NLP window. And since it pretty much spends most of its time screwing up their input upon interpretation and not giving them what they want for even fairly simple cases, it tends not to get used except for very few areas of trivial encyclopedic fact comparison (population of Borneo and the population of Hawaii). In effect turning Alpha into a "comparison engine" not a "computation engine" using facts and data anybody could find just as quickly using Google. For people that have the knowledge to query non-NLP systems, the power of Alpha is locked away behind the sucky NLP interface with no recourse. In effect, it's like putting a Ferrari engine in a combine harvester with a broken steering wheel. So now, the idea is to stick Bing's query engine on top of Alpha's NLP engine, which is on top of their ontology? That's like taking our franken harvester and putting a good car audio system in it as well.