Wizard of OZ Oxford, Maurice Vereecken Goal of implementing Woz • Wp2 – Quick insight in mapping events to Ontdeknet – Working towards what to do in WP3 – Finding the possible obstacles on forehand. • Wp4 – Validating intervention model – Ad-hoc testing of new ideas Application O O O O O O O Response Input Output Implementation;Overview O O O O O O O Parameters þ þ þ ý Intervention þ Decision þ þ ý o o o o o o o Intervention þ Decision þ þ ý Intervention model Atgentive Intervention model Atgentive Step 1; Gather Step 2; Wizard of Oz Step 3; Create modell Step 4 Gather measurable information out of the application and filter them on usability Create a decision list and do some Woz itterations to check the Intervention responses Create the model out of the decision matrix Implement the model in the atgentive interface, and repeat the WoZ itterations to check the model Repeat gathering unitl we have enough to give the right intervention Repeat WoZ until modell refelcts WoZ Output Step 1; Gather information Application O O O O O O O Parameters þ þ þ ý Step 1; Gather Gather measurable information out of the application and filter them on usability Input Response Step 2; Ozzing O O O O O O O o o o o o o o Intervention þ Decision þ þ ý Atgentive Step 2; Wizard of Oz Create a decision list and do some Woz itterations to check the Intervention responses Step 3; Create model Intervention þ Decision þ þ ý Intervention model Step 3; Create model Create the model out of the decision matrix Step 4; Implement model Intervention model Atgentive Step 4 Implement the model in the atgentive interface, and repeat the WoZ itterations to check the model What is woz • The Wizard of Oz technique enables unimplemented technology to be evaluated by using a human to simulate the response of a system. Interacting with the ‘knowledge’ the agent would have. • Actors – Wizard (Observer) – Dorothy (Logged in user) What is woz Using Woz • The "wizard" sits in a back room, observes the user's actions, and simulates the system's responses in realtime. Often users are unaware (until after the experiment) that the system was not real. • The "wizard" has to be able to quickly and accurately discern the user's input. The output must also be sufficiently simple that the "wizard" can simulate or create it in real time. • The easy way: – The user logs in to the system – The observer receives all the application events the system generates – The observer interpretes the data and gives a response based on reasoning (ie; a reasoning tree) Example 1; Introduction • Outline – In the introduction fase, the user has to click on “edit’, fill in his personal introduction and has to click on “save” – The introduction is finished when he has filled in more than 5 words. Example 1; Introduction • Events – When clicking on “edit” a “start task” is fired – When clicking on “save” and more than 5 words are filled in a “complete task” is fired. • Parameters – The context; Introduction – The text; The filled in introduction Example 1; Introduction • Fill in your introduction – The user provides a description/story about himself. Current problem: – Is the user done introducing himself; When is the task finished? Example 1; Introduction Example 1; Introduction Example 2; Assignment • Outline – The user has to open an assignment – Make a decision if we want to use an expert which he is already signed on or search for a new one. – Select the expert for the project Example 2; Assignment • Events – Start_event on opening assignment – Start_event on selecting expert tab • Parameters – The assignment context; Example 2; Assignment • Navigation; (three modalities) – Tell the user; Give a hint how to get to the assignment – Guide the user; tell all the steps the user must take (Buttons etc) – Redirect; Directly show the right screen Example 2; Assignment • Select expert for the assignment – The user can choose one of his own experts or (s)he can sign up with a new one. Current problem: – Signing up with a new expert can be an optional task. How do we manage that? – The user can work parallel on multiple assignments, activities like reading a diary are not linked with one single assignment – The user can click anywhere, the intention is not clear. – If the teacher decides the task isn’t finished, the completed task has to be reopened Example 2; Assignment Example 2; Assignment Next steps • Wp2 – Validate events mapping to model – Creating model on base of lessons learned • Wp4 – Check working of model – Fine tune model on base of pilots