From dialogue design to interaction design

Ann Blandford UCL Interaction Centre at CALRG 09

92 views

Cloud created by:

Gr�inne Conole
18 May 2009

A starting point educationmixed initiative dialogues

Are there educationaladvantages to being able to argue 0 with computer-based tutor about designdecisions?

How might one designan interactive tutor that is able to argue with the student abut designdecisions, does this approach look promising?

The result � WOMBAT amixed initiative system that applied rules to determine what kind of respondingto make to each utterance from the user

Features � dialoguerules that determine which speech acts were appropriate in the current context,stylised sentence construction

Reflection: It worked,did what it was meant to do, but also it didn�t work � no obvious successes,doesn�t actually fit into any obvious educational context

Career path � maths,software engineering, educational technology, agent architectures for dialoguedesign, cognitive modelling, human �computer interaction, studying situatedinteractions and design (distributed cognition, sensemaking, and human error).Shift from formal learning context to more situated learning

Interaction design �puts people at the centre, starts from premise that technology should fitaround the people rather than the converse, need to understand people and howthey are working in order to design effective tools for them

Example � from digitallibrary design to the information journey

People engage with forphases: need, acquisition, interpretation, use

Cycle of informationphases � for example a patient finding out and using information related tosome health issue or students finding information for a course assignment. Informationneed, information acquisition, information interpretation, information use

Supporting sensemakingexamples : interactive visualisations to explore space, richer context tosupport ill-defined searching

Information provisionview contrasts with the lived experience

Resources are beingwasted on inappropriate technologies based on incorrect assumptions about infopractices

�A richer understanding of what people isdoing is needed

Extra content

Got extra information or live blog notes about this cloud? Add them here!

Discussion

Ann Blandford talking about WOMBAT - a system for supporting dialogue developed as part of her PhD. She has now moved on to Interaction design as her focus and showed an example of looking at digital library design as an information journey. The way the need, acquisition and interpretation and use all link together - the e.g. of a patient was compared with students. Students were characterised as "satisficing" based on looking at assessment criteria. The deeper sensemaking information is not supported by Google - new approaches are needed.

Patrick McAndrew
3:22pm 18 May 2009

PhD (1988)

Are there educational advantages to being able to �argue� with a computer-based tutor about design decisions?
Big question, in wisdom of hindsight probably slightly misguided
How might one go about designing a computer-based education tutor that could �argue�? (or at least look like it was arguing, if the same things repeated would come back with the same things - not hyperintelligent, but could argue to a limited degree?)
The result was WOMBAT
Generated set of dialogue rules, determined what speech acts appropriate, sentence construction terribly stylised
Not very good user interface - bit embarrassing now professor of HCI!
Reflecting - worked in that it did what it was meant to do ('verification' test)
But didn't work in that no obvious successors ('validation' test) - resonance with Mike's talk earlier about having to work on different levels.

Current work

Interested in distributed cognition, human error, sense making, situated learning more than formal learning
HCI starts from premise that technology should fit around people rather than the converse, so need to understand the way people work in order to design effective tools
Been working with Anne Adams on digital libraries
Notion of an information journey - people don't just find information and do something with it immediately - co-evolving spiral of phases such as interpretation/sense-making
Worked with variety of populations of users: journalist, lawyers, clinicians and patients, students andstaff
Example : Patients. Often start with recognised problem, either symptom or external event (such as MMR vaccine). Questions like 'Do I need to see a doctor?', 'What will the doctor want to know from me?', after visit, checking the doctor told the truth. Information acquisition best supported (e.g. nhs direct, web) but some sources of dubious quality. How do people interpret and contextualise what they find relative to own situation?
Example 2: Worked with group of undergrads at UCL over several years. Looked at imposed problem of finding information for course work. Rather than doing what most lecturers would like - engage with the problem, students anticipated demands of the assessment, selecting topics they are comfortable with, reinterpreting question as one they felt comfortable with etc. Odd exception but not many. Took very minimal approach to make interpretation as unthreatening and easy as possible.
Example of how people interpret lots of information - lawyers looking at commentaries on the law as well as the law itself, journalists laying out information on their desk, clinicians consulting peers to check interpretations of published infromation
Lots of focus on making information available over the web, but that only satisfies part of what people really need.