I have a quick question about CTAKES.
I am using AE “AggregatePlaintextUMLSProcessor.xml” and want to get some
negation results by referring to polarity attribute.
However, it turns out, for example “Negative for hepatitis”, is not negated. I
think it is weird and I tried “No hepatitis”, “ Denies
Hi Yu,
The new polarity module is machine-learning based so it is not always
easy to diagnose accuracy issues. But generally it might mean there was
no example like that in the training data. It was trained on multiple
corpora, but sometimes certain phrases slip through the cracks, and
"Deny hepat
I think this demonstrates that machine learning is not the right approach
to the negation/polarity problem.
Michael Gurley
m-gur...@northwestern.edu
312 925 3268
Northwestern University Clinical and Translational Sciences Institute
(NUCATS)
http://www.nucats.northwestern.edu
Rubloff Building
750
cTAKES also implements a rule-based approach to the negation/polarity problem.
It was the default until the latest release. You are free to use the rule-based
implementation and compare results with the ML approach.
--Guergana
-Original Message-
From: Michael J Gurley [mailto:m-gur...@no
Hi Michael,
I'm somewhat sympathetic to that opinion. But we did a bunch of
experiments and it seemed to us that negex was too hand-tailored for a
specific dataset and that our new module did better across datasets and
overall. The tradeoff is that it is harder to improve and it sometimes
gives une
Tim, I like your idea of a hybrid approach. I've thought about trying a
hybrid approach in the past myself, but haven't had a chance to try it or
seen any papers on it. It seems you could do it by either treating the
NegEx output simply as a feature in the ML model or combining the output of
NegEx
Hi folks,
Michelle, Sean, Guergana, and Co. have created a few mockups for the new cTAKES
website. Which option would folks prefer?
This is purely on the design intent, and layout, etc. (not actual content).
Option 1: http://mwchen.scripts.mit.edu/cTakes/mock0/index.html
Option 2: http://mwchen.
I prefer option 1. Largely because of the prominence of ‘downloads’ and
‘examples’.
Britt Fitch
Wired Informatics
265 Franklin St Ste 1702
Boston, MA 02110
http://wiredinformatics.com
britt.fi...@wiredinformatics.com
> On Dec 31, 2014, at 2:53 PM, Chen, Pei wrote:
>
> Hi folks,
> Michelle, S
For front page I prefer 1 or 4, for similar reasons as Britt. I love the figure
in option 3 and we should use it but as the first thing you see it is a little
dense. I like the color of option 1 better than option 4.
Tim
On 12/31/2014 03:05 PM, britt fitch wrote:
I prefer option 1. Largely beca
I prefer option 4 overall for compactness, for the prominence of the download
button, and the green color of the button. But of all the bars at the top, I
prefer the look of the top bar from option 2.
I agree with Tim about the figure in option 3.
-- James
Thanks for creating all those mockups! Agree with James and Tim, I prefer
option 4 for its clear layout and the emphasis of ctakes' pros.
Happy New Year!
Best,
Chen
-Original Message-
From: Masanz, James J. [mailto:masanz.ja...@mayo.edu]
Sent: Wednesday, December 31, 2014 3:25 PM
To: de
As I was reading this thread I had the same thought as Tim, perhaps a
combination. It seems over the perfect training corpus this wouldnt be
necessary, but perhaps as a stop gap the "ensemble" approach for some using
your training data but working in a diff corpus (not that I really have the
time t
Acknowledging how variable taste is in aesthetics like web design... I'll
say I really like 1! Though 4 would be a strong contender.
The graphic in 3 is very informative.
JG
On Wed, Dec 31, 2014 at 4:09 PM, Lin, Chen
wrote:
> Thanks for creating all those mockups! Agree with James and Tim, I p
Hi,
>From my s4 loved the first option.
Options 3 and 4 weren't working.
Oleg
On 31 Dec 2014 22:00, "Chen, Pei" wrote:
> Hi folks,
> Michelle, Sean, Guergana, and Co. have created a few mockups for the new
> cTAKES website. Which option would folks prefer?
> This is purely on the design intent,
14 matches
Mail list logo
