[iks-community] IKS workshop "Semantic Search - Fact and Fiction"

[Stéphane Croisier]

Hi John,

At 13:12 04.09.2009, you wrote:
>3. Who are the people doing great things in the 
>area of search within the CMS world that you would like to see at the workshop?
>4. Who would be on your wish list from industry 
>or academia beyond the CMS world to talk about 
>"Semantic Search" at the workshop. Maybe 
>personalities from Google, W3C, Bing, Yahoo, Lucene, Wolfram Alpha?

Regarding point 3 and 4, I would propose to you 
SalsaDev (http://www.salsadev.com/). This is a 
bit biaised as I am one of the Board Member of 
this company (http://www.salsadev.com/team). This 
is a new Swiss start-up which is developing some 
interesting things around the notion of vectors, 
LSA and other similar algorithms in order to 
automatically extract the meaning of any piece of 
information. We were speaking a lot during the 
last event about RDF; OWL; Linked Data & co but 
it would also be interesting to investigate a bit 
more the second facet of the semantic world 
driven by the automatic extraction of concepts 
and the representation of a vector-based 
knowledge map of the information based no more on 
keywords, associations and metadata but on the 
underlying abstracted senses and distances between vectors.

We had an interesting dinner yesterday with the 
SalsaDev folks and Bertrand Delacrétaz. They have 
a nice demo based on a Firefox plug-in which 
could automatically understand the different 
meaning of the web page your are currently 
browsing and can automatically suggest to you 
some related articles (they currently indexed 
wikipedia as a demo example but this could 
perfectly rather be related to articles stored in 
your private CMS repositories instead).

In the same perspective you could also perhaps 
try to invite Recommind 
(http://www.recommind.com/) which is also 
leveraging the PLSA algorithm 
(http://www.recommind.com/technology/core) and 
recently launched their new Categorization 
MindServer in order to automatically categorize 
doucments 
(http://www.cmswatch.com/Trends/1665-Recommind-Auto-categorization). 
Automony and Reuters OpenCalais look like also interesting people to invite.

I could also suggest Zemanta: 
http://www.zemanta.com/  which was not present 
last time and is located in Slovenia.

As everybody looke like interested by a semantic 
Lucene, it would be interesting to invite the 
newly created Lucid Imagination company which 
regroups some of the key committers of the Apache 
Lucene project (6 mio USD raised from the CIA end 
of last 
spring:http://www.cmswire.com/cms/enterprise-20/cia-invests-in-open-source-lucene-solr-search-004830.php). 
There is the ApacheCon US 2009 
(http://us.apachecon.com/c/acus2009/) the week 
before in California but perhaps they will be 
interested to come or to send someone. However I 
do not know if they really have any semantic 
initiatives for the moment. But perhaps it could 
be interesting to have them on board and help 
make some contacts between them and other 
Lucene-based projects. Another interesting Apache 
project currently under incubation is UIMA 
(<http://incubator.apache.org/uima/>http://incubator.apache.org/uima). 
I personally do not know any of the current 
committers 
(http://incubator.apache.org/uima/team-list.html) 
but it would be interesting to also get them on board.

In France, one of our customers (Manpower: 
http://www.cio-online.com/actualites/lire-manpower-procede-a-l-analyse-semantique-des-cv-non-structures-2353.html) 
integrated Lingway. They perhaps also have some 
interesting demo and use cases to show to us: 
http://www.lingway.com/index.php?lang=en

Some thoughts....
Stéphane

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.iks-project.eu/pipermail/iks-community/attachments/20090904/2c6dcd4c/attachment.htm>