Research #53

opinionfinder

Added by Thanh Tam Nguyen almost 2 years ago. Updated over 1 year ago.

Status:Closed Start:2008-10-11
Priority:Normal Due date:2008-11-05
Assigned to:Thanh Tam Nguyen % Done:

0%
Category:- Spent time: -
Target version:- Estimated time:10.00 hours

Description

OpinionFider is an open source software which automatically processes documents and identifies subjective sentence as well as polarity. It can also identify subjective expressions and speech events. Here is the latest release http://www.cs.pitt.edu/mpqa/opinionfinderrelease/

Installing OpinionFinder is a little bit difficult. It can only work on Linux and use a lot of other library such as WordNet, OpenNLP, BoosTexter... We also have to open the source code to fix some error before compiling it.

After processing document, OpinionFinder returns, for example:

<MPQASENT autoclass1="unknown" autoclass2="obj" diff="3.8">We have no record of Mark Twain's earliest letters.</MPQASENT> <MPQASENT autoclass1="subj" autoclass2="subj" diff="29.2">Very likely

&lt;MPQASRC&gt;they&lt;/MPQASRC&gt; were soiled pencil notes, written to some school sweetheart--
to "Becky Thatcher," perhaps--and tossed across at lucky moments,
or otherwise, with &lt;MPQAPOL autoclass="positive"&gt;happy&lt;/MPQAPOL&gt; or &lt;MPQAPOL autoclass="negative"&gt;disastrous&lt;/MPQAPOL&gt; results.&lt;/MPQASENT&gt;  &lt;MPQASENT autoclass1="unknown" autoclass2="subj" diff="32.1"&gt;One of those
smudgy, much-folded school notes of the Tom Sawyer period would be
priceless to-day, and somewhere among forgotten keepsakes it may
exist, but we shall not be likely to find it.&lt;/MPQASENT&gt;  &lt;MPQASENT autoclass1="unknown" autoclass2="obj" diff="9.1"&gt;No letter of his
boyhood, no scrap of his earlier writing, has come to light except
his penciled name, SAM CLEMENS, laboriously inscribed on the inside
of a small worn purse that once held his &lt;MPQAPOL autoclass="negative"&gt;meager&lt;/MPQAPOL&gt;, almost non-existent
wealth.&lt;/MPQASENT&gt;  &lt;MPQASENT autoclass1="unknown" autoclass2="subj" diff="7.4"&gt;He became a printer's apprentice at twelve, but as he
received no salary, the need of a purse could not have been urgent.&lt;/MPQASENT&gt;
&lt;MPQASENT autoclass1="subj" autoclass2="subj" diff="25.5"&gt;He must have carried it pretty steadily, however, from its
appearance--as a kind of symbol of &lt;MPQAPOL autoclass="positive"&gt;hope&lt;/MPQAPOL&gt;, maybe--a token of that
Sellers-optimism which dominated his early life, and was never
entirely subdued.&lt;/MPQASENT&gt;

History

Updated by Kuiyu Chang almost 2 years ago

thanks for the update, you can install opinionfinder on liubei, ssh into liubei via

ssh -p 9673

if you need any software/modules installed on liubei, please let me know.

Updated by Kuiyu Chang almost 2 years ago

  • Due date set to 2008-10-30

Continue to install and compile opinion finder on liubei, let me know if you need any other software installed.

Updated by Thanh Tam Nguyen almost 2 years ago

OpinionFinder has been installed in /home/soft/opinionfinder so far. To test it, please go to '/home/soft/opinionfinder' and use following command:

python opinionfinder.py -f database/docs/twain.doclist

Where twain.doclist is the file containing a list of documents

The output data should be in database/output_anns folder

Updated by Kuiyu Chang almost 2 years ago

  • Due date changed from 2008-10-30 to 2008-11-05
  • Estimated time set to 10.00

Run opinion finder on the annotated (for just 1 chosen query like "N95") dataset of 50-100 files, and compute the baseline precision and recall.

This will serve as a reference.

Updated by Kuiyu Chang over 1 year ago

what's the status on your personal annotation engine?

Updated by Thanh Tam Nguyen over 1 year ago

  • Status changed from New to Closed

After considering my personal annotation engine and GATE again, I decided to work on GATE for annotating.

Also available in: Atom PDF