## Are you using Wolfram Alpha?

Hi!

That blog-post will not be very long, because I want to sleep soon and do further investigations later.

Many of you will know about Wolfram Alpha, when I first saw it, I thought: hey, that is really cool. Many people are using it. But it has some short comings:

• They want you to buy Mathematica
• It is proprietary
• You cannot create arbitrarily complicated queries
• You cannot save temporary results
• Sometimes English grammar is too ambiguous
• No localization

Well, that service should not be used, it is actually not that great, and like Google Mail, Chrome OS, iOS or WP7 it is becoming really popular and is a threat to Free Software in my opinion, especially high qualified people are using it and stop caring about Free Software (“just a web-app”). But what are the alternatives?

• There is Sage – an awesome computer algebra system combining a lot of free software, using Python instead of ugly, specialized languages like in Mathematica, Maple and Maxima
• Semantic databases in RDF-format, e.g. DBPedia crawling the Wikipedia or governmental websites like data.gov or data.gov.uk

For people being able to use SPARQL the mentioned RDF-databases are very powerful tools – no limitations on queries or something like that. So what is missing? A nice interface anybody can use. It would be really nice to have a free tool parsing natural language queries (e.g. using Earley-parsers and some probabilistics, it would not really have to understand real sentences, just some fixed structures would be enough like “where”, “by”, “all” and fixed sets of attributes). Those queries could be transformed into e.g. DBPedia-SPARQL-queries, and the RDF-results could be transformed into some nice tables and maybe graphs. A free implementation would be a) awesome and b) much more seriously usable than Wolfram Alpha. Do you know about any project trying to do something like that? Any comments? What is your opinion about such software?

Regards

PS:
A few random thoughts about it:

• Of course natural language stuff would be nice for math, too.
• With SPARQL access from Sage/Python data-querying and calculations would also be combinable. A software could generate Python containing SPARQL.
• I do not know if there are practically usable distributed RDF-databases, but such software could make it possible to distribute query evaluation to the peers, Free Software projects probably cannot afford Wolfram’s computing-capacity.
• Combination of different RDF-databases may be a problemtask.

### 14 Responses to “Are you using Wolfram Alpha?”

1. awer Says:

If it works I don’t care if the data is here or there, anyway, I write only something like 6x=2, not big deal.
Anyway, the future seems to be in the web, and I don’t have to buy & install the program.

2. JR Says:

I used yesterday WA to calculate past currency rates when doing some booking.

“120000 ISK in SEK at 14 February 2010″

Sure, I wouldn’t mind using an open source alternative, and I could probably write an SQL query to fetch that if I knew the names of the fields. But the ability to write your query in dumbed-down English is a strong point.

3. The User Says:

@awer
So you do not care about freedom of software and knowledge? I think for web-applications it is as important as for desktop-applications. It does not matter if everybody would use a free webbrowser and a free kernel, but dozens of proprietary webservices to do their work. The post was primarily about the data fetching, not stuff like “6x=2”, but as I said – also for all those calculations a free alternative usable by anybody would be important.

@JR
Of course that is a strong feature, and that is why I am asking for alternatives.

4. ruurd Says:

In short – it is a failure. If they would have taken a good look at Google they would not have made these errors. As so many of Wolfram’s stuff: of little interest. Academically interesting maybe but hampered by bad business decisions.

5. The User Says:

@ruurd
What do you mean? I know really many people using it, nobody is using free alternatives… And I do not get your point, what is better about Google? What should they do like Google?

6. Trurl Says:

Proprietary software is no threat for free software. PEOPLE are a threat. The users must and will decide: if a proprietary software is better than a free software, the proprietary software is in some sense rightfully used and the free software must improve. And vice versa, of course. Without competition, things would get worse over time, not better.

Thus quoth Trurl.

7. The User Says:

About boring SAP-spam-software you may say: who cares? (though there are different arguments in that case) But behind software like Wolfram Alpha there are relevant technology, mathematics etc. which should be public for the benefit of science and society. But if everybody is using the proprietary stuff, Free Software and their algorithms etc. will not evolve and people will depend on proprietary services by companies like Wolfram.

8. CSousa Says:

Sage is awesome awesome awesome!

We can even run it online: http://www.sagenb.org/ (slow as hell, but we can use other faster servers).

9. Framp Says:

Well, wolfram alpha is pretty much the best free (and closed) resource available.

I mainly used wolfram alpha for its calculus explanation feature (which isn’t available in Mathematica 8).
There is no free software I’m aware of which can do that.
To be fair, I still have to find a free software which is able to solve complex calculus query

10. The User Says:

@Framp
Played a bit around, Sage (Maxima) is pretty good at solving differential equations and integrals, could not find anything Wolfram Alpha could solve, but Maxima could not. The usage of those explanations is quite limited, e.g. weird substitutions etc., of course it may be helpful, but it is not worth it to use such proprietary stuff. It is not slower and it is more likely to tell me that he cannot solve it (Wolfram Alpha does not want to tell me quite often and the progress-bars are moving on ;)). Try it out. It is Python, it is very nice. A natural language interface is needed in my opinion to attract students etc.

PS:
In very simplistic cases the steps by Wolfram Alpha are useful, but for complex cases you will not get anything. Sage will at least tell you which algorithm it used.

PPS:
Wolfram Alpha stupid result for differential equation, Sage:

desolve(diff(y,x)**3+x==y(x),y,ivar=x,show_method=True,contrib_ode=True)

Output:

[[[x == 3/2*t^2 + c + 3*t + 3*log(t - 1), y(x) == t^3 + x]], 'lagrange']

Could still not find anything Wolfram could solve but not Sage.

11. maninalift Says:

This is a pretty big project, on the other hand it is something free software could get very right since it involves creating a good abstraction then adding many modules for specialist knowledge to it.

12. g Says:

@The User: maybe you can tell me how to calculate in Sage what in WolframAlpha can be calculated with “integrate y * (cos x)^5 dx dy from x=sqrt(y) to 2 from y=0 to 4″? In Maple the symbolic result of this integral can be found, I am still looking for a way to get the symbolic result in Sage. How do you calculate in Sage “eigenvectors {{0.03,0.03,0.03},{0.48,0.03,0.93},{0.48,0.93,0.03}}”?

13. The User Says:

Quite simple, just remember that in Sage variables are not automatically symbolic variables, they are Python-variables and they need a value (pi, e, i etc. are default constants, and x is default symbolic variable):

y = var('y')
assume(y >= 0)
integral(integral(y*cos(x)^5, (x, sqrt(y), 2)), (y, 0, 4))

It works, and it is Python, yeah!

14. g Says:

Ah, thanks, I had forgotten the variable declaration for y. The fact that x is already a symbolic variable (and the fact that in Maple you don’t have to declare symbolic variables) always makes me forget to declare the other variables.

XHTML: Use <blockquote cite="name"> for quotations, <pre lang="text    ∨ cpp-qt ∨ cpp ∨ bash ∨ other language"> for code, [latex] for formulas and <em> for em. Contact me if the comment does not get published, it may have accidentally been marked as spam.

Anti-Spam Quiz: