|
On 2/7/13 11:34 AM, Ali SH wrote:
Hi Kingsley,
Ok, let me see if I've understood you - so you provide 3
links comparing the search functionality of Siri to the search
functionality of Google Search, where Google ... unsurprisingly
performs better. What exactly does this prove? To me it hints at
non-understanding of what Siri's primary application is.
Sorry, but that's an incorrect assessment. I am not knew (in any
way) to Siri.
What's more confusing is that in the "apples-to-apples"
comparison, you suggest that Watson, a deep QA system that is
aimed at Fortune 500 company budgets is the appropriate
comparison to a smart phone app that you used to be able to
download for $0.99. This is an apples-to-apples
comparison???
You asked about LOD (a constellation of data sources) in comparison
to Siri (an application). Watson made/makes us of LOD. Siri's use is
a mystery. My fundamental issue is that Siri (if it was more loosely
coupled) should be able to enrich its knowledge via access to
knowledgebases from subject matter experts. That's what you get from
Linked Data and the LOD cloud. The loosely coupling is implicit by
virtue of URIs, entity relationship semantics, and inference etc..
A couple of points...
First, I think it's become clear that Apple has lobotomized
Siri.
Yes, and that's my fundamental point!
There are a variety of articles that attest to this, and the
fact that some of the key developers of the Siri technology have
left Apple seems to suggest there may have been a significant
warping of that technology.
That said, I think it's important to emphasize that it's main
functionality has never been search, but to perform
transactions.
I never claimed that. It is supposed to provide answers to questions
via a voice interface.
And it does (did) so, but really interpreting the transaction
components of target websites / services in a machine-readable
way, rolling it all out in one app... I think this is something
that we often forget, and I'm grateful for John Sowa and Ronald
Stamper for bringing semiotics back into the picture. It really
helps clarify a lot of the confusion that arises from simply
thinking of semantics divorced from pragmatics...
Yes.
In any event, from my perspective, Siri comes a lot closer to
the original SW vision of making a lot of the human web, machine
readable.
It's a poor demo of what's actually possible today. If it used what
was possible today it wouldn't exit out to Web searches across
Wikipedia. It would be able to perform the kind of disambiguation
that would flip the script (Judo style) on Google, but that isn't
the case due to Apple's choices.
They were able to use structured data where it was provided,
otherwise they did the hard work of analyzing what was available
at a given site / service, curating the information and mapping
it to their domain and transactional ontologies.
That's where the like for like comparison comes in with Watson. They
did that too, but made smarter use of the LOD cloud datasets.
In the end, the Siri team chose to wall-off their
interpretation, which imo is a business decision that we can
argue about, though it certainly gave them a first mover
advantage (and led to the founding team earning a nice $250M
payout)...
Remember, wealth isn't the basis of worth to me. They build a silo
that's imploded, predictably.
With regard to your specific comparisons --- well, Siri is
not intended as primarily a search tool.
And what is Google then?
It was developed to be a transactional tool, and I don't see
how Google Search performing better than Siri proves anything.
It proves that Google (partial Semantic Web exploitation) can easily
trump Siri (which claims to be a Semantic Web exploitation
showcase). If Siri implemented their system more like Watson, they
wouldn't be the case.
The Androi-Siri videso and page you link to, simply shows
the testers comparing the performance of the two apps based on a
variety of search based questions (many of which are outside of
the Siri domain, where it degrades to search)...
Performance and Scalability are features. They are integral
components of Google's competitive advantage.
Kingsley
On Thu, Feb 7, 2013 at 6:19 AM,
Kingsley Idehen <kidehen@xxxxxxxxxxxxxx>
wrote:
On 2/6/13 11:59 PM, Ali SH wrote:
Hi Kingsley,
You make some sweeping claims here...
On Sun, Jan 27, 2013 at
7:20 PM, Kingsley Idehen <kidehen@xxxxxxxxxxxxxx>
wrote:
Yes, and the tragedy of Siri is that
lightweight semantics applied to
unstructured data runs rings around it, as
demonstrated by Google.
Look, if the Siri folks (many who actually
know better) had implemented this solution
with a little more selflessness they would be
running rings around Google today. You
wouldn't have Google making a mockery of Siri
on its own home turf.
No it doesn't. The Semantic Web isn't about silo
applications. It's all about the Web and the addition of
entity relationship semantics to the mesh of content
constitutes the Web. If Siri was even marginally close to
the essence of the Semantic Web vision it would have loose
coupling to data sources.
Loose coupling to semantically rich data sources data
sources on the Web is how Siri would have protected itself
from what's happening right now i.e., Google runs rings
around it, playfully.
In search? Yet that is not where its prime competency
is... I fail to see the relevance.
That they didn't expose their
own methods via open standards, is a business
decision, but i don't think that's their "downfall"
(if we can even call it that). It is true that Apple
to a large extent lobotomized Siri, but I can't make
head or tails of your claims.
Siri is like a human being today that doesn't have the
ability to access new knowledge from subject matter
experts. It just knows what it knows and its sources are
mysteriously limited.
Since it was initially developed to perform transactions
on behalf of people, it has a very carefully selected and
curated set of competencies, since trust is an absolute must
in this domain. When Apple bought the technology, they
certainly curtailed its abilities while simultaneously
marketed it with a lot of smoke and hype.
In the end, it seems that most of the reviews are
fundamentally not understanding what the technology is
about.
In what ways is Google making a "mockery" of
Siri?
There are many videos demonstrating that [1][2][3].
I don't think a single one demonstrates that. It only
does so if you misunderstand the intent of the technology.
It does show that Google performs better on search though...
Google has a virtuous cycle the works well for its goals.
It isn't pure, but it is superior to Apples' approach.
Ok, on this point I agree. Apple has certainly warped
Siri.
Is there a single SemWeb or LOD project that
comes close to even a fraction of Siri's
functionality? C.f:
Compare like with like. Maybe you mean: is there a similar
smart agent that leverages the LOD cloud that one could
compare with Siri. My basic answer: IBM Watson [4].
How is this a fair comparison? A technology aimed at
companies with budgets in the millions vs an app aimed at
individuals with a budget of $0.99
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
--
Regards,
Kingsley Idehen
Founder & CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca handle: @kidehen
Google+ Profile: https://plus.google.com/112399767740508618350/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen
|
smime.p7s
Description: S/MIME Cryptographic Signature
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J (01)
|