Hard-coded paths in medfacts/cleartk

[Thien Nguyen <ng...@gmail.com>]

Hi all,

It's been awhile (some of you may remember me from ARC) but I'm working on
embedding the latest cTAKES 3.2.2 into some tooling. I ran into an issue
with classes in the `org.apache.ctakes.assertion.medfacts.cleartk` because
the analysis engines there have explicit paths for the
`ContextWordWindowExtractor`. For example:

    public class HistoryCleartkAnalysisEngine extends
        AssertionCleartkAnalysisEngine {
        ...
        private void initialize_history_extractor() {

            if(this.entityFeatureExtractors == null){
                this.entityFeatureExtractors = new ArrayList<>();
            }
            this.entityFeatureExtractors.add(*new
ContextWordWindowExtractor("org/apache/ctakes/assertion/models/history.txt")*
);
            this.entityFeatureExtractors.add(new
HistoryFeaturesExtractor());
        }
        ...

And this forced me to put those files into the root of my project (i.e.
/my-project/org/apache/ctakes/assertion/models/history.txt). Since it seems
all other resources have the paths configurable in the descriptor xml
files, should these follow suit and be considered a bug?

Thanks again for all your work on cTAKES. Fun to watch it go from
pre-Apache days to where it's at now.

Thien

Re: Hard-coded paths in medfacts/cleartk

[Pei Chen <pe...@wiredinformatics.com>]

Hi Nguyen,
Good to hear from you?  How is ARC doing?  I was looking to download a copy the other day but couldn't find it online anymore. 

But yes, we can make those paths configurable just like the models. 
If you can, open a Jira to track this? (Anyone should be able to create a new Jira account)

Sent from my iPhone

> On Oct 15, 2015, at 2:55 PM, Thien Nguyen <ng...@gmail.com> wrote:
> 
> Hi all,
> 
> It's been awhile (some of you may remember me from ARC) but I'm working on
> embedding the latest cTAKES 3.2.2 into some tooling. I ran into an issue
> with classes in the `org.apache.ctakes.assertion.medfacts.cleartk` because
> the analysis engines there have explicit paths for the
> `ContextWordWindowExtractor`. For example:
> 
>    public class HistoryCleartkAnalysisEngine extends
>        AssertionCleartkAnalysisEngine {
>        ...
>        private void initialize_history_extractor() {
> 
>            if(this.entityFeatureExtractors == null){
>                this.entityFeatureExtractors = new ArrayList<>();
>            }
>            this.entityFeatureExtractors.add(*new
> ContextWordWindowExtractor("org/apache/ctakes/assertion/models/history.txt")*
> );
>            this.entityFeatureExtractors.add(new
> HistoryFeaturesExtractor());
>        }
>        ...
> 
> And this forced me to put those files into the root of my project (i.e.
> /my-project/org/apache/ctakes/assertion/models/history.txt). Since it seems
> all other resources have the paths configurable in the descriptor xml
> files, should these follow suit and be considered a bug?
> 
> Thanks again for all your work on cTAKES. Fun to watch it go from
> pre-Apache days to where it's at now.
> 
> Thien