If you have "no place to go," come here!

Volunteers wanted for project -- trade deals? -- using keen annotation tool to which I have access

Egas was originally designed for annotating biomedical texts with conceptual relations, but the nice thing is that you can develop your own types of relations, not just the ones they supply. Here's an academic review of the EGAS platform:

Egas is a web-based platform for biomedical [but can be anything!!] text mining and collaborative curation. It allows users to annotate texts with occurrences of concepts and relations between these concepts. The annotation tool follows what we termed an “annotation-as-a-service” paradigm. Document collections [for example, of trade treaties], users, configurations, annotations, back-end data storage, as well as the tools for document processing and text mining, are all managed centrally. This way, a curation team can use the service, configured according to their requisites, taking advantage of a centrally managed pipeline.

The tool is based on the idea of Projects. A Project consists of a curation or document annotation task, performed on a collection of documents, by a team of (one or more) curators, and considering a pre-defined set of concept and relation types defined by the curation guidelines. A project administrator is responsible for managing the users (curators) associated to the project and the project characteristics, such as annotation guidelines and target concepts and relations. Projects may be public or private, in which case they are only accessible by users that have been added by the project manager.

My thought was that this would be an ideal medium to help many readers sense [*** cough *** mellon *** cough ***] of the trade deals. You could load up the project with a corpus of documents, develop a set of relations, then annotate the corpus to show the concepts and relations implicit in the text, allow the use to filter on relations, etc. This is really keen stuff.

Go play; it's biomedical data, but I hope the potential is clear. Here's a screen shot:

To me, this is pretty thrilling; and I might add that I to a degree, and much more letsgetidone, have professional-level expertise designing concepts and relationships for large corpora (not that I can volunteer lets).

So we have an opportunity here to do some interesting and useful work. Could be any kind of document, but I figured that trade was hot right now, and somebody's going to want to look at the Wikileaks stuff in broader context.

egas.jpg116.92 KB
No votes yet


Submitted by lambert on

As we used to say in the software business, "with a level of effort," this could become a very useful tool, because the trade agreements are -- by design -- so tangled and obfuscated. My thought would be to focus on the trans-national sovreignty-destroying relations, since that is the hot button....

letsgetitdone's picture
Submitted by letsgetitdone on

But I certainly think the effort is a great one.

In the field of knowledge management, there a great deal of disagreement about how "knowledge" should be defined? It's a problem in epistemology that philosophers since the Greeks haven't been able to get a consensus answer to.

My answer was given here. By all means read the piece. But long story short in the context of using the EGAS platform. Its use by many people in creating annotations on the trade deals will create one kind of knowledge I identify. Specifically a kind of knowledge defined as:

. . . tested, evaluated, and surviving, sharable (objective), linguistic formulations about the world (i.e., claims and meta-claims that are speech- or artifact-based or cultural knowledge used in learning, thinking, and acting).

Note the "meta-claims" in the definition. In most texts the claims are there, but the "meta-claims" are not called out and recorded in any useful structure. So, it's hard to pick out the knowledge in the text, or to look at the track record of claims relative to meta-claims over many texts. What you're proposing here is to use EGAS to create meta-claims in the form of annotations. So the result would be "objective knowledge." That doesn't make ir right or "true." But it does make criticizable and testable over time, and that's what makes knowledge claims, "knowledge."

Submitted by lambert on

.... but and so I had more though of asking you to backstop the taxonomy rather than to day to day stuff. An email exchange or two, not more :-)

Actually, I built a similar system in Drupal but never deployed it; it was based on Wiki-like markup and I felt it was too hard to use... Perhaps I should look at it again....