- From: Stan Devitt <stan.devitt@gwi-ag.com>
- Date: Wed, 7 Jun 2006 09:42:03 +0200
- To: 'Bijan Parsia' <bparsia@isr.umd.edu>
- Cc: 'Alex Kozlenkov' <alex.kozlenkov@betfair.com>, Gerd Wagner <wagnerg@tu-cottbus.de>, public-rif-wg@w3.org
For an representative example of the kind of thinking prior to the start of the first MathML working group look at http://www.ucc.ie/info/net/html/htmlmath.announce The MathML working group came out of a desire to go a bit beyond (or instead of) that kind of approach and make embedded math formulas accessible to computation engines, but with no real plan beyond possibly adopting one CA vendors notation and asking browsers vendors to parse strings in a non-xml syntax and add rendering with something of the complexity of TeX. As for success and wide adoption going way beyond such hopes: 1. adding over a 1000 characters added to unicode. 2. vendor work leading to mechanisms for direct inclusion and display in a number of browsers including ie6 and mozilla. (a major hurdle) 3. direct mathml support added to several xml editors, all major Computer algebra and numeric systems. 4. incorporation of mathml into the production process for major math publishers, and into electronic course delivery systems such as WebCT and BlackBoard. 5. adoptions by math related societies and also the patent office which alone generates over 12000 pages a month. 6. inclusion of a mathml core set of dictionaries in OpenMath. 7. Recent experimental extensions to Arabic Math without major design flaws being encountered and strongly underscoring that there is a many to many relationship between notations and concepts. 8. Impact on other working groups as a difficult test case for various technologies. The main places I am seeing dissatisfaction is from two areas. At one end we have those who want to type 2D^2y on their html page and have it display like mathematics and at the other we have those who want to incorporate a more formal approach to definitions and types. The former is currently only addressed by preprocessing into MathML, while the latter is addressed by referencing more formal structures and associating it with a visual display. It was a conscious design decision to go this way. We deliberately opted for preprocessing of strings as it directly conflicted with the goal of archiving semantics (there is no standard meaning to such an expression, even to the level of assuming multiplication or standard meanings for (say) D). We ruled out the latter as existing proposals were too experimental and increased the complexity of an already complex proposal. We felt there was enough in place to experiment with the concepts and standardize later and provided examples in Notes of how a systematic use of the existing structure could address some concerns. ---- As for the goals of the working group and their refinement, the group was driven largely by: 1. There could only be one MathML. 2. We had to be able to render Math to the quality of TeX. 3. We had to be able to archive computable expressions in a manner which captured their computational semantics. 4. Not everybody was interested in both 2 and 3 but both were essential. 5. meeting requirements 2 and 3 often led to different syntactic structures. Everyone on the working group worked to achieve those goals including making them work together, even though many were primarily interested in only one or the other -- and largely because of a recognition of 1. By analogy I would argue that while RIF members may feel that they are only interested in one formalizm there is an obligation to see if we can provide a framework that accommodates them. Stan -----Original Message----- From: Bijan Parsia [mailto:bparsia@isr.umd.edu] Sent: Tuesday, June 06, 2006 5:51 PM To: Stan Devitt Cc: 'Alex Kozlenkov'; Gerd Wagner; public-rif-wg@w3.org Subject: Re: "industry needs" On Jun 6, 2006, at 4:29 PM, Stan Devitt wrote: > The point here was not to debate the details and perceptions of of > MathML I wasn't really trying to debate it, I was merely trying to learn from that experience. I agree there are analogies between RIF and MathML, but I also am trying to nail down what "widespread adoption" really means, esp. as a *general* CSF. But it's hard to get very far from your elliptical remarks, hence the request for further points. I merely included the other pointers as evidence that I did try to find this out myself (however quickly). > though I think it and the process that was followed does provide good > counter examples to some of the ideas I have been hearing. [snip] Not if they aren't accessible. Or rather, it's harder for them to do so. Of course, this being the W3C, the discussions, at least, are open. But the adoption probably comes from other sources. I tend to think that "widespread adoption" is a bit of a empty whatever it is (requirement?). (In fact, I had an email from someone off list where they said they understood "widespread adoption" to mean "widespread adoption in the 'mainstream' rules community" where the "'mainstream' rules community" was dominated by PR rules. So, "PR rules are essential to widespread adoption" becomes, well, a tautology.) I tend to think it's more important to know if *members* are going to adoption/support it. So Alex's report is much more interesting and informative (my skepticism wrt to the 900 IT engineers notwithstanding) than these other ones. Forgive my skepticism, but when people appeal to certain facts, I prefer some back up for those facts, particularly about things that are in principle verifiable. Cheers, Bijan.
Received on Wednesday, 7 June 2006 07:42:28 UTC