Annotating information in tables with WebAnno

[Martin Lentschat]

Helo everyone,

I am currently working on a framework for annotating experimental data in scientific papers. My first case study is the annotation of information related to food packaging permeabilities (e.g. permeability measurements, packaging component, control parameters...).

I am currently struggling with the annotation of the tables. I manage to keep the structure of the tables in the .txt files I use for the import. However I am runnig into two problems:

(1) the ammount of relations created makes a mess of the text (see my screenshots bellow)

(2) at some point, when creating a new relation, I need to reload the page to display it.

Before annotation:

After annotation:

(1) is the most important issue, as it just makes the annotation work extremely difficult. As you can see it is pretty unreadable.

(2) is probably related to the hosting condition of my server (i use the docker version of WebAnno and it is hosted on an university server).

So I wanted to ask if any of you have already performed this kind of tasks and if you had any solution or advice for me.

Thanks you all for your help !

Martin

[Richard Eckart de Castilho]

Hi,

> On 19. May 2020, at 14:35, Martin Lentschat <martinl...@gmail.com> wrote:
>
> I am currently struggling with the annotation of the tables. I manage to keep the structure of the tables in the .txt files I use for the import. However I am runnig into two problems:
> (1) the ammount of relations created makes a mess of the text (see my screenshots bellow)

You might get a slightly better experience enabling the "collapse arcs" option in the settings dialog
on the annotation page.

That said: maybe it would also work for you to act without the relation entirely:

- you already annotate the measure_Unit - I would add a feature to that annotation which says which unit it is exactly, (e.g. Gramm, Pascal, etc.)
- then you could also a unit feature to the numeric_Value annotation with the same set of values (Gramm, Pascal, etc.)
- that gives you the knowledge that both annotations refer to the same measurement type - even without the relation

Maybe that is sufficient and you do not need the exact coreference information provided by an explicit relation.

Cheers,

-- Richard

[Martin Lentschat]

Thank you Richard for your realy quick answer.

I do not have any "collapse arcs" option in the settings. Is it a new option ? (my version is 3.5.5.)

As for the addition of features on the annotation, it is an option we already explored with my colleagues. The number of variations is too large for this to be a viable solution, as it replaces the trouble of drawing the relations with entering a valid feature.

The information in tables is not the center of our work right now. We will probably use the .html version to parse the content, probably with just the annotation the table heads.I mostly wanted to know if you ran into this kind of tasks already.

Thanks again for your response and your work on that great tool.

[richard...@gmail.com]

try the latest WebAnno 3.6.x Release.

As for the tables you might give the pdf support in INCEpTION a try. No promises though. Let me know If you try how you fare.

— Richard

Sent from my mobile, sorry for brevity.

[Martin Lentschat]

Hi Richard,

I tried importing a pdf into INCEpTION demo server (pdfTEST project) and noticed two main problems:

(1) large numbers of tockens are not discriminated against. This results in the merging of many words (e.g., "atheend" instead of "at the end").

(2) the tables appear on a single line, making annotation extremely difficult.

The PDF version of my files also contains a lot of unwanted text (like the publication information). This is more of a personal problem. I use the .html version of the articles to build a .txt adapted to WebAnno.

Martin,

Le mardi 19 mai 2020 14:35:05 UTC+2, Martin Lentschat a écrit :

[Richard Eckart de Castilho]

Hi,

thanks for the feedback.

> On 20. May 2020, at 10:33, Martin Lentschat <martinl...@gmail.com> wrote:
>
> I tried importing a pdf into INCEpTION demo server (pdfTEST project) and noticed two main problems:
> (1) large numbers of tockens are not discriminated against. This results in the merging of many words (e.g., "atheend" instead of "at the end").

This could be worked around by configuring the annotation layers to anchor on characters instead of tokens.

> (2) the tables appear on a single line, making annotation extremely difficult.

Did you change the visualization on the annotation page from "brat" to "pdf"? This can be done via the settings dialog on the annotation page.
If you do that, you see the PDF and can directly annotate on the PDF. How well this works depends on the particular PDF files.

-- Richard