Using a Turk Talk scenario

Turk Talk is a new function we have built into OpenLabyrinth to provide a hybrid approach to natural language processing (NLP). Rather than rely on very sophisticated NLP algorithms, which are still somewhat in their infancy, we instead use human facilitators to briefly interact with learners in small group work.

The overall approach is described more broadly here:

If you are interested in trying this functionality, here are some notes on how to set things up. You will need to use Scenarios in OpenLabyrinth to do this. Scenarios are a way to group together case series, along with specific user groups.

First, design your map so as to include some Turk Talk questions. We strongly suggest that you don’t overuse this function – it is easy to overload your facilitators or Turkers. Choose a 2-3 spots within your case where free text responses to a question are important. For the rest of the interactions, continue to use the more standard means of interaction such as MCQs, drag & drop questions, links and dandelions. For simple free text NLP, you can even use Rules to provide some basic text parsing.

A Turk Talk question is designed like any other in the Question Editor. Choose a stem, the dialog box parameters etc. Note that when a Turker is not present, there is no way for OpenLabyrinth to respond. Be mindful of this in your map designs and allow for a by-pass of a Turk Talk question for when testing if the Turker is not present.

Once you are happy with your map design, attach the map to a Scenario step. Also attach to that Scenario those Users who you want to monitor as learners. You need to have a user status of Director or Superuser to manage Scenarios.

For more information on how to use Turk Talk scenarios: