Moses-support Digest, Vol 102, Issue 35

Send Moses-support mailing list submissions to
moses-support@mit.edu

To subscribe or unsubscribe via the World Wide Web, visit
http://mailman.mit.edu/mailman/listinfo/moses-support
or, via email, send a message with subject or body 'help' to
moses-support-request@mit.edu

You can reach the person managing the list at
moses-support-owner@mit.edu

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Moses-support digest..."


Today's Topics:

1. Re: [decoding-graph-backoff] (Matthias Huck)
2. Research position in La Rochelle, France : event detection,
sentiment analysis (Antoine Doucet)


----------------------------------------------------------------------

Message: 1
Date: Thu, 16 Apr 2015 22:26:41 +0100
From: Matthias Huck <mhuck@inf.ed.ac.uk>
Subject: Re: [Moses-support] [decoding-graph-backoff]
To: Hieu Hoang <hieuhoang@gmail.com>
Cc: moses-support <moses-support@mit.edu>
Message-ID: <1429219601.30904.54.camel@portedgar>
Content-Type: text/plain; charset="UTF-8"

I think your remark in the mail from January was correct, it has to be
ePos-sPos+1 > backoff
but currently still is
ePos-sPos+1 <= backoff

Are you able to somehow test this?


On Thu, 2015-04-16 at 23:57 +0400, Hieu Hoang wrote:
> ah yes, I thought the backoff was doing the opposite to what it's
> supposed to do so I changed the comparison around. I checked that it
> backed off, but i didn't run it through tuning.
>
> it may still be wrong, or there may be strange interaction with the tuning.
>
>
> On 16/04/2015 22:16, Matthias Huck wrote:
> > Well, what's that business mentioned in your mail from January (quoted
> > below), with the backoff being broken, then being broken more, then
> > possibly been fixed - or not?
> >
> > https://github.com/moses-smt/mosesdecoder/commit/44fec57c535db2df73ccbb1628d8143a9c728c19
> >
> >
> > I set up a system that was supposed to do backoff with factored
> > generation steps, more or less in the manner of what's described in this
> > paper: Interpolated Backoff for Factored Translation Models, Philipp
> > Koehn and Barry Haddow, AMTA 2012.
> >
> > MIRA tunes all the weights of the backoff models to 0. With exactly the
> > same configuration, this did not happen last year (February 2014). Maybe
> > the [decoding-graph-backoff] setting didn't have any effect prior to
> > some of your code modifications, and the models were actually competing
> > in older setups? Or it's buggy now. I can't really tell.
> >
> > I can show you the two setups if you want.
> >
> >
> >
> > On Thu, 2015-04-16 at 21:34 +0400, Hieu Hoang wrote:
> >> Didn't know it has changed. How should it behave and how does it
> >> actually behave?
> >>
> >> On 16 Apr 2015 21:04, "Matthias Huck" <mhuck@inf.ed.ac.uk> wrote:
> >> Hi Hieu,
> >>
> >> It seems that [decoding-graph-backoff] doesn't quite behave
> >> like last
> >> year any more. Can you briefly explain how its behaviour has
> >> changed,
> >> i.e. what it did before and what it does now? Can you please
> >> also let me
> >> know whether there's a way to reproduce the old behaviour via
> >> configuration options?
> >>
> >> Cheers,
> >> Matthias
> >>
> >>
> >>
> >> On Fri, 2015-01-09 at 15:20 +0000, Hieu Hoang wrote:
> >> > >From the git history, I think it was slightly broken, then
> >> I broke it even
> >> > more in May 2014.
> >> >
> >> >
> >> https://github.com/moses-smt/mosesdecoder/commit/44fec57c535db2df73ccbb1628d8143a9c728c19
> >> >
> >> > It was
> >> > endPos-startPos+1 >= backoff
> >> > then
> >> > endPos-startPos+1 <= backoff
> >> > I think it should be
> >> > endPos-startPos+1 > backoff
> >> >
> >> > I'll change it if it's ok with everyone
> >> >
> >> >
> >> > On 9 January 2015 at 15:11, Marcin Junczys-Dowmunt
> >> <junczys@amu.edu.pl>
> >> > wrote:
> >> >
> >> > > Hm, we have been using it at WIPO, but I have to admit I
> >> never checked
> >> > > it _actually_ does anything useful. We sorta believe it
> >> does.
> >> > >
> >> > > W dniu 09.01.2015 o 16:08, Hieu Hoang pisze:
> >> > >
> >> > > Hi All
> >> > >
> >> > > Does anyone use this functionality in Moses when you have
> >> multiple
> >> > > phrase-tables?
> >> > >
> >> > > From the code, it doesn't look like it works as described
> >> in
> >> > > http://www.statmt.org/moses/?n=Moses.AdvancedFeatures
> >> > >
> >> > > Maybe I'm missing something
> >> > >
> >> > > --
> >> > > Hieu Hoang
> >> > > Research Associate
> >> > > University of Edinburgh
> >> > > http://www.hoang.co.uk/hieu
> >> > >
> >> > >
> >> > >
> >> > > _______________________________________________
> >> > > Moses-support mailing
> >> listMoses-support@mit.eduhttp://mailman.mit.edu/mailman/listinfo/moses-support
> >> > >
> >> > >
> >> > >
> >> > > _______________________________________________
> >> > > Moses-support mailing list
> >> > > Moses-support@mit.edu
> >> > > http://mailman.mit.edu/mailman/listinfo/moses-support
> >> > >
> >> > >
> >> >
> >> >
> >> > _______________________________________________
> >> > Moses-support mailing list
> >> > Moses-support@mit.edu
> >> > http://mailman.mit.edu/mailman/listinfo/moses-support
> >> > The University of Edinburgh is a charitable body, registered
> >> in
> >> > Scotland, with registration number SC005336.
> >>
> >>
> >>
> >> --
> >> The University of Edinburgh is a charitable body, registered
> >> in
> >> Scotland, with registration number SC005336.
> >>
> >>
> >
> >
>



--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.



------------------------------

Message: 2
Date: Thu, 02 Apr 2015 12:40:01 +0200
From: Antoine Doucet <antoine.doucet@univ-lr.fr>
Subject: [Moses-support] Research position in La Rochelle, France :
event detection, sentiment analysis
To: destinataires inconnus:;
Message-ID: <551D1C81.5080209@univ-lr.fr>
Content-Type: text/plain; charset="utf-8"

Please find below the announcement of a postdoctoral position in the
University of La Rochelle, France, on the topics of touristic event
extraction and sentiment analysis.

If you have a Ph.D in computer science and you have working knowledge in
the fields of NLP, IR, Big data and/or ontologies, and if you like the
idea of joining a dynamic research group that lies within walking
distances of 3 different beaches, please let us know!

If you have trouble viewing the description below, please visit the
following url:
http://pageperso.univ-lr.fr/antoine.doucet/PostDoc_EventDetection.pdf

Please note the short deadline: 14 April 2015!!

(apologies for cross-posting)


***Post?Doctoral Position*


*Sentiment Analysis and **touristic event **extraction*


The L3i laboratory, within the Tourinflux project (http
<http://www.tourinflux.com/>:// <http://www.tourinflux.com/>www
<http://www.tourinflux.com/>. <http://www.tourinflux.com/>tourinflux
<http://www.tourinflux.com/>. <http://www.tourinflux.com/>com
<http://www.tourinflux.com/>/ <http://www.tourinflux.com/>) is seeking a
postdoctoral researcher in computer science, on the topic of sentiment
analysis and touristic event extraction.


/Length: 10months/

/Expected recruitment: 1^st of June 2015 or ASAP ? the position is to be
filled ASAP, and will last until the end of February 2016./

/Net salary: 2100 ? monthly/

/Location: L3i laboratory, La Rochelle, France/

/Fields: Computer Science / Natural Language Processing / Semantic Web/

/Keywords: E-tourism, Multilingual NLP, Event extraction, opinion
mining, sentiment analysis, NLP, Normalization(schema.org,
TourInFrance), Semantic Web, Data analysis, spatiotemporal events./


*Job description:*

The work will be conducted in the informatics, image and interaction
laboratory (L3i), within the Tourinflux project, funded by a public
investment program for the future(PIA). The L3i is a 120-person
laboratory created in 1993. Hosted in the historical and sunny city of
La Rochelle (http://www.holidays-la-rochelle.co.uk/), it is ranked A by
the French research evaluation agency (AERES).


In addition to the L3i, the Tourinflux project involves 2 companies and
theprofessional association for the digital economy, in collaboration
with several actors of the French tourism industry. The project aims at
providing actors of the touristindustry with a set of tools allowing
them to handle both their internal data, and the information available
on the Web, so as to better diagnose and influence on the perception of
territories. The tools currently available are insufficient, and a lot
of the data gathering, analysis and processing is currently done by
hand, or via the use of various tools that are only partially
satisfactory. Tourinflux aims to provide an extensive dashboard,
allowing all institutions, whatever their size, to visualize and
interpretthe information available about their territory. This should
allow them to improve their decision process and subsequently, their
effectiveness.


Specifically, this postdoctoral position will focus on the automated
extraction of touristic events and/or opinion mining over touristic
objects. Both subjects will be considered on equal groundsand the
ability of the candidate will be the main recruitment criteria. The
techniques to be used will need to be applicable to any language.


1.

*(**Semi**)**-automated **extraction **of tourist**ic****events**:*

Tourism information is both heterogeneous (free text, Web page,
pictures, ?) and semi-structured. Extracting touristicinformation is a
major challenge at a time when the masses of unstructured information
are constantly evolving, be it on the Web or within organizations.The
task of the recruited postdoctoral researcher will beto extract
touristic events and insert information into a given representation
model. Two key challenges were identified:

*

Adapting NLP techniques to the specific domain of tourism
information, and extract the key features of the field

*

Understanding and feeding the given models of representations for
touristic data.


Recent work on the temporal analysis of natural language, so as to grasp
the opening hours of touristic objects remains to beintegrated. It will
be a plus if the recruited postdoctoral researcher is able to integrate
this work within the event extraction process.


2.

*Sentiment analysis **over touristic objects*

The postdoctoral researcher will be in charge of extracting sentiment
from online data, e.g, evaluation platforms (such as tripadvisor),
blogs, microblogs, ? The extracted data is to be adapated to our own
opinionated mark-up language (see, eg, SentiML). This task will require
three steps:

*

Given a touristic object, fetch online information about it;

*

Extract sentiment about the object;

*

Fill our annotation schema.


*Specific requirements:*

Candidates must have PhD in computer science, with abilities in
knowledge representation and data mining. Research experience is also
required in at least 2 of the following domains:

*

Natural Language Processing, Text mining

*

Information Retrieval

*

Big Data and Data Warehouses (eg, Hadoop)

*

Modeling, Ontologies and inference engines

*

Annotation and Evaluation methodologies


*General requirements:*

*

One or more of the following programming languages: Python, C/C++, java?

*

Team-work skills (knowledge of Agile methodologies would be a plus)

*

Proven ability for scientific writing


*To apply:*

All candidates are required to send a resume, an expression of interest,
and the names and contact information of at least 2 references
(including email addresses) to:

*mickael.coustaty@univ-lr.fr*

*antoine.doucet@univ-lr.fr*


_*Please note that applications must be sent at the latest on
*__*Tues*__*day *__*14*__**__*April *__*at 23:59 CET*_


--
Antoine Doucet
Full Professor
L3i - Laboratoire Informatique, Image et Interaction
University of La Rochelle - IUT de La Rochelle
http://pageperso.univ-lr.fr/antoine.doucet/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/mailman/private/moses-support/attachments/20150402/60633616/attachment.htm
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Image1
Type: image/png
Size: 2818 bytes
Desc: not available
Url : http://mailman.mit.edu/mailman/private/moses-support/attachments/20150402/60633616/attachment.png
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 1
Type: image/png
Size: 5121 bytes
Desc: not available
Url : http://mailman.mit.edu/mailman/private/moses-support/attachments/20150402/60633616/attachment-0001.png

------------------------------

_______________________________________________
Moses-support mailing list
Moses-support@mit.edu
http://mailman.mit.edu/mailman/listinfo/moses-support


End of Moses-support Digest, Vol 102, Issue 35
**********************************************

Tweet