MathGroup Archive 2009

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Wolfram|Alpha Lookup Tool for Mathematica

  • To: mathgroup at
  • Subject: [mg100107] Re: Wolfram|Alpha Lookup Tool for Mathematica
  • From: dbreiss at
  • Date: Mon, 25 May 2009 06:14:50 -0400 (EDT)
  • References: <gv7r1a$as5$>

On May 22, 11:37 pm, Bill Rowe <readn... at> wrote:
> On 5/22/09 at 1:46 AM, dbre... at wrote:
> >On May 21, 12:05 am, Bill Rowe <readn... at> wrote:
> >>On 5/20/09 at 5:02 AM, mcmcc... at (mark mcclure) wrote:
> >>Entry of search terms in the tool cause your default web browser to
> >>launch and send a query to Wolfram|Alpha. And the tool only allows one
> >>search at a time to be done. The only thing David's tool does is
> >>create a convenient means to start a query while using Mathematica.
> >>From the perspective of the Wolfram|Alpha server's the result is
> >>identical to a human entering a single query in a conventional web
> >>browser.
> >>Consequently, it is far from obvious this is against the terms of use.
> >>And since the query sent to the server is identical in both cases,
> >>there is no way to Wolfram to distinguish between the two cases.
> >This is an interesting point.  I do think that the Terms of service
> >are slightly confusing on this point.  They say  on the one hand
> >"be used only by a human being using a conventional web browser to
> >manually enter queries one at a time"
> >but later on allow one to use hyperlinks from a web page (or I
> >assume, god forbid, a microsoft word document).   So these two
> >things seem in conflict.
> >My tool would appear to violate the terms of service according to
> >the first principle, and would be on the borderline of the second
> >one. (Again, though, I cleared it with the WRI legal department).
> >One question that I want to further clarify is how one can, in fact,
> >use Mathematica to send queries to W|A.  The functions behind my
> >tool (yes, you can reverse engineer it if you know what you are
> >doing... the code is not too deeply encoded) can be used more
> >generally.  And I am thinking of how to extend it.  For example to
> >open several W|A web pages at a time--perhaps you have a list of
> >chemical compounds that you need information of for your ongoing
> >work.
> Assuming Wolfram|Alpha becomes and remains a significant tool to
> get information, I would think WRI will at some point create a
> version of Mathematica with a built-in means to query
> Wolfram|Alpha directly.
> >Of course , from Mathematica, you could send a command to your
> >browser to open a hundred web pages at a time.  This would be very
> >bad.
> And is undoubtedly one of the main reasons for stating searches
> need to be entered manually by a human.
> >And, in the case of W|A it would probably cause the W|A servers to
> >block your Internet IP address.  So, anything that one creates needs t=
> >be built with care.
> >For example in my tool, it limits the length of the query to 128
> >characters, so that you don't accidentally paste the entire Bible
> >into the input-field and try to send that to W|A as a query.  This
> >respects the rough actual size of the W|A website's input-field.
> Which is why I would argue your tool complies with the terms of
> service. Unless I modify your tool, as designed it only allows
> one query at a time which has to be manually entered. And by
> restricting the length of the input to be the same as the W|A
> query box, in effect, tool is no different whatever from
> entering the same query in a box displayed by a web browser.

I think that this is indeed generally true and would be the way that
WRI would view it.  I actually designed the tool the way I did because
it made sense to do it that way, given the thing (W|A) that it was
interfacing with.  (And, in fact, I wanted to make sure that no user
of it would accidentally get in trouble from using it....   )

But there are open questions concerning writing functions in
Mathematica  that reach out to W|A.  And I am forwarding some of them
that I have come up with to the WRI legal folks.

I suspect that the terms of service is an evolving document (just as
the W|A business model is) and so exploring new ways to use W|A will
raise new questions that hadn't been thought of previously.  The terms
of service would likely then be modified once the issues (legal and
business) have been thought through.

For those who are curious, I actually discussed the tool with the WRI
legal team originally because I wanted to be sure that I was not
violating any trademark rights.  But letting them know about it also
covered the other potential issues.

>  The only technical difference I see between your tool and entry into
> the box displayed by a web browser is that I don't literally
> enter the query into the box displayed by the web browser. And
> since the actual query sent to W|A is identical in both cases,
> there will be no way for WRI to determine what interface I used
> to enter the query on my machine. Nor is there any reason for
> them to care.
> The reality is for ordinary users the terms of service are
> essentially meaningless. It is not reasonable to think all or
> even most users of W|A will have read the terms of service and
> understood them in any significant way. And as you point out
> above, if a user creates problems there are ways of dealing with
> them regardless of the terms of service. Also, the process for
> dealing with an abusive user will not be significantly different
> whether the terms of service existed or not.

  • Prev by Date: Bug of NotebookRead
  • Next by Date: Re: Accessing Dictionary definitions
  • Previous by thread: Re: Wolfram|Alpha Lookup Tool for Mathematica
  • Next by thread: 2D ray tracer