There is a small correction with respect to the percentage value given to
the entity. I previously thought that the percentage value with the entity
indicated how sure OpenCalais was on the element being an entity. That turns
out to be incorrect.
Thomas from OpenCalais team mailed me the correction and it goes like this:
Just a quick note on an email I saw on this list. The % returned by Calais
with an entity is not the certainty of correctness – but something much
cooler. It’s actually the computed relevance of the entity to the document
as a whole. Feed it an article with a bunch of entities mentions and we’ll
attempt to identify the most important entities in the content.
Great for noise level reduction in tagging, etc.
Thanks Thomas.
On Mon, Oct 19, 2009 at 8:24 PM, Arvind Jamuna Dixit ard...@gmail.comwrote:
The demo given by Anand on semantic search of feeds using OpenCalais was
very interesting.
The fact that his app essentially did something that
Feedlyhttp://www.feedly.com/( based on Google Reader ) does piqued my
interest.
Feedly too uses OpenCalais to get semantic information on a blog post and
for search.
If you use Google Reader as a RSS reader then you have to check out Feedly.
Anand used python-calais http://code.google.com/p/python-calais/ for
communicating to the OpenCalais
APIhttp://opencalais.com/documentation/calais-web-service-api/api-invocation/rest
.
Anand then showed how OpenCalais categorized the input text into entities,
topics and relations.
With every entity OpenCalais attached a percentage that indicated how sure
it was on the element being the entity
whose identity was given by a URL. For instance, there is a URL identiying
the entity Android as a product.
Open Calais is exciting new technology that has made information gathering
applications like Feedly and Klezio http://www.klezio.com/ possible.
Another application is Tagaroo http://tagaroo.opencalais.com/ a
wordpress plugin that automatically generates tags as you type your blog
post.
In the meetup Anand said that the current state-of-the-art of semantic web
is not even 5% of its potential. A quick search of the term
imagine semantic
webhttp://www.readwriteweb.com/fastsearch?search=imagine+semantic+webx=0y=0on
ReadWriteWeb shows that a lot of businesses and visionaries are betting
big on semantic web and know that
it has a lot of potential.
On Mon, Oct 19, 2009 at 11:17 AM, Anand Balachandran Pillai
abpil...@gmail.com wrote:
On Sun, Oct 18, 2009 at 9:30 PM, Sriram Narayanan sriram...@gmail.comwrote:
On Sun, Oct 18, 2009 at 8:52 PM, Noufal Ibrahim nou...@gmail.com
wrote:
Hi,
I had some stuff to take care of and was too tired to attend the
meeting. How did it go? Any tidbits that the list might benefit from?
The opencalais session was good. I'm going to tell my colleagues about
this, and they'd surely explore it more.
Anand and Ramdas have asked that we have sessions again with more
detail. I agree, since the scant amount of ZFS that I showcased today
was in itself radically different from what we all know from other
file systems.
I request feedback from other attendees on today's ZFS session. Please
let me know what else you'd have liked me to cover at an introduction
level.
We had a good meeting. There were some initial hiccups with the
projector but moving to a separate meeting room solved it.
I presented a brief introduction to semantic web and showcased the
OpenCalais API using python-calais. I showed how to extract semantic
concepts (categories) from existing data using the API.
I went on to demo my application which listens to a couple of mobile
phone news feeds and uses semantic information returned by OpenCalais
to provide specific natural language queries (not exactly NLP there yet,
but I am simulating NLP like queries) which return specific answers.
The demo showed making a query on cost of motorola android and
this returning the specific data requested on the price of the most
recent motorola android mobile phones.
I will wait for some other attendee for their feedback on how good
this was rather than making the comment myself :)
This was followed by a very good session by Sriram and Moinak on
the capabilities of ZFS. I don't want to get into details, but I was blown
away by the capabilities of ZFS. I had only read about it before and never
seen it in action, so when Sriram showed how to increase the storage
of an existing volume by adding another device and just adding it to
the volume using zfs add it was just too good to believe. ZFS
makes those actions which could take hours using Linux ext3
look trivial and done within seconds...!
We started the ZFS session a bit late i.e around 5.30 pm so there
was not much time to showcase all the bits planned. We dispersed
around 6.10 pm.
We were 7 attendees in total.
-- Sriram
___
BangPypers mailing list
