Tutorial 3: Automated tagging and hand correction in the UI

In Tutorial 2, you used a command-line tool to build a model. In this step, we'll use the UI to process documents using, among other things, the model you built. We're going to use the same simple named entity task that comes with MAT, and we're going to assume that your task is installed.  Like Tutorial 2, we're going to do this tutorial in file mode.

In this tutorial, we're going to make use of the model we built in Tutorial 2.

Step 1: Start up the Web server and UI

See the section on starting the UI in Tutorial 1. If you failed to shut down your Web server at the end of Tutorial 1, shut it down and restart it now; any models you built in Tutorial 2 won't be visible to the server until you restart it.

Step 2: Prepare a document for automated tagging

Back in Tutorial 1, we used the UI to prepare a document for hand tagging. Please repeat steps 3 and 4 from Tutorial 1. Once you complete those steps, your window should look like it did at the end of step 4:

[voa2 zoned and toked]

Note that the zone and tokenize steps are grayed out, indicating that they've been completed, and that the word units in the document are delimited by gray boxes.

Step 3: Tag

You may recall from Tutorial 2 that we built a default model for the Named Entity task. The UI knows about this default model, and we can use it now.

Press the gear button. The "tag" label in the Status line should blink momentarily while the Web server loads the model, and then you should see a window like this:

[voa2 zoned, toked, tagged]

The gear is now grayed out, and the "tag" step is partially grayed out, indicating the step is partially done, and hand annotation is available if you want to correct the output of the automated tagger.

Step 4: Undo and redo

As we pointed out in Tutorial 1, this workflow permits you to undo an operation. Let's say, for instance, you've applied the model as in step 3, but you haven't done any hand correction, and you've subsequently prepared a more accurate default model, and you want to undo the automated tagging and apply the newer model.

Press the backward button. The UI will warn you that it's about to undo unsaved changes:

Press "Yes", and the document will return to the state it was in at the end of Step 2. Press the forward button again, and the document will return to the state it was in at the end of Step 3.

Step 5: Hand correct and complete

At this point, you can hand-correct annotations in the same way you added and edited annotations in Tutorial 1. You can also save the document, as you did there. Once you've completed your hand correction, you can press the forward button (the one with the drawing hand on it). The UI should look something like this now:

[zone, tok, tag, done]

Hand annotation is no longer available, the "tag" step is now completely grayed out, and the buttons have changed once again. Now, we have two back buttons. The one on the left undoes the "tag" step, which will remove all the annotations; the one on the right will retreat back into the hand annotation phase of the "tag" step. It will not undo any annotations; it will simply mark the document as not yet complete, and give you the chance to do additional hand annotation.

Step 6: Clean up (optional)

If you're not planning on doing any other tutorials, and you don't want the "Named Entity" task hanging around, remove it as shown in the final step of Tutorial 1.

This concludes Tutorial 3.