Hi everyone,

Well, I'm finally back from the Wikimania conference in Alexandria. I had a
great time: I met a lot of interesting people (some of whom are on this
list), saw some very informative talks, gave a talk of my own, and got to
discuss SMW-related issues with various people. And of course, there was
Egypt, a fascinating country I'd never been to before. I'll reserve this
email for just observations related to Semantic Forms and Semantic
MediaWiki; if you have any interest in reading about the rest of my
experiences at the conference and in Egypt, I'll probably post some things
to my blog later.

First of all, there was my own talk, titled "Creating the structured
semantic wiki", which I thought went well. It came on the third day of the
conference, and was placed right after the SMW talk by Markus (that one was
originally structured as a panel featuring Markus, Denny and me, but it
became a more standard description-of-SMW presentation, which was most
likely for the best, given the general lack of knowledge about the
extension). The two presentations somewhat merged into one, with overlapping
topics and many people staying to hear both. Thanks to a fantastic setup at
the Library of Alexandria, all the talks at the conference were videotaped;
most of them aren't online yet, but hopefully they will be soon, at
webcast.bibalex.org (and then afterwards, of course, at
techpresentations.org).

There were a few high-level Wikimedia people there for Markus' talk, which
may or may not have meant something; less for my talk, partly, I think,
because I was competing with a presentation by Florence Devouard, who was
about to end her tenure as Wikimedia Foundation president. One person who
did stay for my talk was Angela Beesley, who has connections to both Wikia,
the wiki-hosting company, and Wikimedia; afterwards, she asked me about
Referata, the hosting site I run, which I had mentioned at the end, and how
I viewed it vis-a-vis Wikia (Wikia now offers support for SMW, though not
for any of the other semantic extensions). I told her that I don't view the
two as competitors, since Wikia is intended for general-use, large-scale
encyclopedias, while Referata is intended more for directories and corporate
usage, and I encouraged her to add support for SF and the other semantic
extensions to Wikia, and also to put in a good word for doing the same thing
on Wikipedia itself. We'll see if anything comes out of either of those.

The conference was also memorable for its cultural exchange with the people
of Egypt and the Arab world, who made up a large number of participants; one
interesting anecdote was that, of all the questions I was asked after my
presentation, the one I thought was most insightful was posed by a female
Egyptian student, wearing a headscarf; she asked whether it was a problem
that things are referred to by name in the system, and thus that having
different values with the same name would lead to bad data. "Er, yes, that's
definitely a problem," I responded. I noted that there were a few potential
solutions to it, and said that if I'd created a slide title "Drawbacks of
semantic wikis" (there was one titled "Benefits of semantic wikis"), that
certainly would have appeared on the list. Anyway, it was an interesting
experience.

Earlier, there was a very informative talk by Brion Vibber, head MediaWiki
developer, entitled "State of the MediaWiki". Among other things, Brion
listed changes that were coming to MW and Wikipedia. Among the additions
that he discussed were a new extension, called "CommentPages", that will
allow for standard blog-style comment pages (some people have asked before
about SF being used for such a purpose); improvements to the LiquidThreads
discussion extension (same); and, maybe most relevant to SF, an dimprovement
to MW's file-uploading capability to make it more user-friendly, including
allowing uploading of multiple files at the same time. He also said that he
didn't think WYSIWYG editing was coming any time soon to Wikipedia, because
he didn't think any of the existing solutions were adequate for handling
templates and other complex structures. During the Q&A session I asked Brion
if, were he to start developing MediaWiki today from scratch, he would still
use PHP. He said yes, which I found reassuring, and gave an explanation I
found convincing, which amounted to (a) it makes it easy for people to run
MediaWiki on their own servers, and (b) "it works" (i.e., don't fix it if
it's not broken).

After the talk, a few people came up to Brion to ask some further questions,
and Markus was among them, making the case for including SMW on Wikipedia;
and asking for a clearer protocol for the evaluation of extensions to be
used on Wikipedia. Brion, in his usual Buddha-like way, nodded and gave a
somewhat noncomittal answer to both... alas.

I also met, and saw the presentation by, Michael Dale, who created the
MetaVid extension, and runs MetaVidWiki (metvid.ucsc.edu/wiki), the wiki
that uses it. MetaVid stores second-by-second information about online
videos, including transcripts of what's being said, and it actually uses SMW
to store that information, which I hadn't fully realized before (as I told
Michael, it confuses us when there's no "Semantic" at the beginning of the
extension name). He said that there was a chance that MetaVidWiki would
switch over to using SF for entering semantic information, instead of having
users type in SMW's property tags directly. He's also working on a project
that he's getting funded by the Wikimedia Foundation to do, which also
involves Kaltura, a company that creates software for online video editing.
Basically, I think the idea is to create a single MediaWiki extension that
lets users edit and annotate a video (in the open-source OGG format)
entirely through the wiki; and then eventually get this extension onto
Wikipedia. I asked Michael if this extension could be a "Trojan horse" for
getting SMW onto Wikipedia, but he wasn't sure what the technical details
would be. The founders of Kaltura were also there at the conference, and I
talked briefly to one of them, who said that they might have interest
themselves in using the semantic extensions. We'll see what comes of that.

There was a very interesting talk by Gerard Meijssen, the creator of an
extension called OmegaWiki, which allows for the translation of text phrases
within a wiki, using an interface similar to the one used by BetaWiki for
the translation of phrases used by extensions. The idea of
internationalizing SMW- and SF-based wikis, to show the same set of data
across many different languages, has been brought up before, and this could
be a good fit for that kind of approach.

I also met and talked with Merrick Schaefer, who works at UNICEF and runs a
project called UNIWIKI, which includes both technical and non-technical
aspects, but the technical part is a set of patches and extensions to
MediaWiki that are meant to improve the user interface in order to make it
easier to add pages and view the full set of information; you can see an
example at x.mepemepe.com/index.php/Special:CreatePage . They take a
non-semantic approach, but it could be that the two projects can learn from
each other as far as making user-friendly interfaces.

I also saw a presentation by Mikel Maron about OpenStreetMap, a project to
create freely-available street maps of the entire world using MediaWiki. He
noted some problems with regular web street maps, like Google's, that OSM
seeks to overcome: they're slow to update to changes in the real world, you
can't reuse their data without a license, and they're often not cheap to
use. OSM seems like a natural fit for street mapping with the Semantic
Layers extension, and the only drawback appears to be that their set of data
is currently very spotty outside of Europe, the U.K. and the U.S.

Finally, I had a number of conversations with Markus and Denny outside of
the two presentations; this conference was the first time we had talked in
person about SMW development, so we had a lot to go over. There were two
relevant things we talked about: the first was that I brought up the issue
of SMW supporting true n-ary relations, meaning properties that offer
complete flexibility in setting the number, type and layout of their
sub-properties (see http://semanticweb.org/wiki/N-ary_relations for much
more on this). N-ary relations are relevant to SF for basically any data
that's represented using multiple-instance templates. I think I convinced
Markus and Denny to a greater degree of the usefulness of such a feature,
although the implementation is still yet to be decided, and they said they
didn't have time to add such a thing themselves. The second issue was the
bundling of SMW and the related extensions into a single package. At some
point last year, there was discussion of adding SF and possibly some other
extensions into SMW itself: now the movement is in the other direction, to
make each extension as small as possible in order to increase modularity
(again, with the goal of being added to Wikipedia and other large wikis in
mind). In fact, some components of SMW might themselves get spun off into
separate extensions, like the timeline feature (and possibly any n-ary
relations support that gets added). However, at the same time there's a
desire to bundle all these extensions into a single package, for easier
download, and to synchronize releases to a greater degree. Right now the
plan is to use the "SMW+" project (
http://wiki.ontoprise.de/ontoprisewiki/index.php/SMW%2B_-_Business_Re...)
for this purpose: it's a package created by Ontoprise (makers of the Halo
extension) that already holds a number of extensions (including Halo, of
course), and can have more added to it. This is pending the fixing of Halo
to work with the current version of SMW and be easier to install and
uninstall, which Markus and Denny are now being paid to do.

Well, that turned out to be quite long; and it still didn't cover all the
non-SMW-related stuff from Wikimania, and the non-conference parts... but
hopefully that sums up most of the technical things I learned.

-Yaron