Great talk within HCID Centre seminar series:
Title: Reasoning from Users' Experiences: from Case-Based Reasoning to Traces-Based Reasoning.
Amélie Cordier from the University of Lyon.
Abstract:
Case-Based Reasoning is an artificial intelligence paradigm that consist in reusing previous experiences (cases) to solve new problems. The problem with Case-Based Reasoning is that "cases" are too restrictive to effectively represent "experiences". Traces are relatively new digital objects that enables us to capture "interactions" between users and systems. For numerous reasons, traces are rich objects. Trace-Based Reasoning is an evolution of the Case-Based Reasoning paradigm where cases are replaced by traces. Trace-Based Reasoning can be used for many purposes. One of them is user assistance. This will be the main topic of this talk. Indeed, in this presentation, I will give an introduction to Trace-Based Reasoning and I will show the benefits of Trace-Based Reasoning for user assistance. In the meantime, I will discuss open research questions related to Trace-Based Reasoning. Next, I will give several examples of Trace-Based applications developed in the Silex team (both academic and industrial applications).
Biography:
Amélie Cordier is an associate professor in the University of Lyon. She conducts her research in the LIRIS Laboratory. She received her PhD in 2008. She worked on an interactive and opportunistic approach to acquire knowledge for Case-Based Reasoning systems. Ever since, she works on experience sharing and experience reusing. She is involved in several academic projects (Ideal, Kolflow, Ozalid), and has a couple of partnerships withs industrials (SAP-Bo, IBM). She is interested in Case-Based Reasoning, Trace-Based Reasoning, Users' Experiences, Semantic Web and Knowledge Engineering.
Many interesting case studies online:
Friday, 9 September 2011
Reasoning from Users' HCID Seminar: Experiences: from Case-Based Reasoning to Traces-Based Reasoning
Markers:
Centre for HCI Design,
City University,
data visualization,
HCID seminar,
seminar,
Talk,
trace
Friday, 12 August 2011
Focus Group on Web App for Universities
I was invited this Friday to help my friend Raj Arjan (Interaction Lab Manager) to facilitate a focus group about a Web App for Universities. Took pictures, repositioned the cameras as the process went forward, as well.
Based on student and academic needs, the University has decided that they will provide full access to the information on smart phones and tablet computers. The web site will be turned into a web application that will bring students, employees, and alumni closer together. The new solution will allow everyone access to the information from everywhere without being tied to a desktop or laptop computer.
The focus group was guided through the following steps:
1. Present the video
2. Present the power point presentation
3. Talk briefly about affinity diagramming after the presentation
Then start the group work.
Object clustering task:
Use full white board for this. Produce a list of objects. Put all on the
white board immediately. When they run out of objects, start the
categorisation og them. Objects are yellow, categories are orange, 20
mins. Take a picture at the end of the whiteboard.
Subject clustering task:
Split the white board in two sides: Allowed and Disallowed. Produce a list
of subjects. For each identified subject, categorise the identified
objects on either the Allowed or Disallowed sides of the whiteboard. Take
picture of the whiteboard when they start on a new subject.
Context clustering task:
Split in the white board in two sides: Allowed and Disallowed. Make a list
of contexts. Choose one context and one subject (a context-subject pair).
Put the objects on the Allowed or Disallowed side. Take picture of the
whiteboard when they start on a new context-subject pair.
Based on student and academic needs, the University has decided that they will provide full access to the information on smart phones and tablet computers. The web site will be turned into a web application that will bring students, employees, and alumni closer together. The new solution will allow everyone access to the information from everywhere without being tied to a desktop or laptop computer.
The focus group was guided through the following steps:
1. Present the video
2. Present the power point presentation
3. Talk briefly about affinity diagramming after the presentation
Then start the group work.
Object clustering task:
Use full white board for this. Produce a list of objects. Put all on the
white board immediately. When they run out of objects, start the
categorisation og them. Objects are yellow, categories are orange, 20
mins. Take a picture at the end of the whiteboard.
Subject clustering task:
Split the white board in two sides: Allowed and Disallowed. Produce a list
of subjects. For each identified subject, categorise the identified
objects on either the Allowed or Disallowed sides of the whiteboard. Take
picture of the whiteboard when they start on a new subject.
Context clustering task:
Split in the white board in two sides: Allowed and Disallowed. Make a list
of contexts. Choose one context and one subject (a context-subject pair).
Put the objects on the Allowed or Disallowed side. Take picture of the
whiteboard when they start on a new context-subject pair.
Markers:
focus group,
Interaction Lab,
rajiv arjan,
web apps
Friday, 5 August 2011
New gadgets at the Interaction Lab!
Raj Arjan, my friend and Interaction Lab manager purchased more equipment that will allow more research and interdisciplinary research:
1. Mind control two devices (emotiv and mindwave) mindwave is mac compatible
2. X60 now can track gaming
Still no app for mac besides the sdk
Stand to track mobile and ipad
3. Mini wearable camera, non intrusive
4. Nvidia 3d glasses
5. Galvanic skin response for temperature, motion, skin condutivity - stress levels
UT of gesture based interactions
6. 2 iPad2 tablet computers
7. Didital pad Acecad, handwriting notes being digitalized and drawings/scribbles
8. Panasonic IP camera, remote controlled, controlled by browser
9. Gaming console: nintendo Wii (controller hacks) and x-box360 with kinect (for geatural interaction)
10. Dell laptop to run Tobii eyetracker so we could analyze
1. Mind control two devices (emotiv and mindwave) mindwave is mac compatible
2. X60 now can track gaming
Still no app for mac besides the sdk
Stand to track mobile and ipad
3. Mini wearable camera, non intrusive
4. Nvidia 3d glasses
5. Galvanic skin response for temperature, motion, skin condutivity - stress levels
UT of gesture based interactions
6. 2 iPad2 tablet computers
7. Didital pad Acecad, handwriting notes being digitalized and drawings/scribbles
8. Panasonic IP camera, remote controlled, controlled by browser
9. Gaming console: nintendo Wii (controller hacks) and x-box360 with kinect (for geatural interaction)
10. Dell laptop to run Tobii eyetracker so we could analyze
Markers:
HCI Design Centre,
Interaction Lab,
rajiv arjan,
tobii
Wednesday, 29 June 2011
Some thinking...
Some reviews:
1. The possibility to set up a machine as an active Eye Gaze - using our Tobii X120 (I'm checking the tech specs, George).
2. Reviewing my timeline (next steps until the end).
3. need to write the 1 year report.4
4. Defining the prototype, reviewing objectives, hypotesis, variables, methodology (this is a subject that is coming back in papaer revies and doc consortioums that I was declined...too 'high level' is the general comment).
5. Updates about my qualitative analysis on the web 2 drag and drop experiment.
Remember: Metrics by Fixation Duration, Total Fixation duration, fixation count, fix count including zeros
Very interested on what happens when one comes across a novel form of interface, specially Post-WIMP with NUI modes of interaction
Very interested on the experience of what happens when one is offered an interaction with multiple inputs (gesture, eye gaze, voice) and with novel GUI solutions.
Present more graphics to justify my point!
Present more percentages within the problem. How many r interested? Any info online?
It's happening. But I should run a questionnaire if people r interested on engaging on Post-WIMP NUI
Studies about linguistics/natural gestures/gaze commands for the new interface
That was not done before. What am I improving? What novelty am I bringing?
Am I developing a technique/protocol? Recommendations for design?
Add a slide with completed work so far with checks (eg bibliographic review)
Check success/failure rate bibliography
Tool to capture websites: captures AJAX-style/SSL/Data mining (web mining)
RECOGNIZE is better then RECALL
how to analyze user's memory? What theory should I use? Within Cogn Psy?
How to analyze behaviour? Gestures language, grammatics, shall I bring back semiotics, linguistics
Define properly my variables
Correct my doc consortium doc
Declare general and specific obj
FRAMEWORK FOR EVALUATION
What is the Scenario?
1. The possibility to set up a machine as an active Eye Gaze - using our Tobii X120 (I'm checking the tech specs, George).
2. Reviewing my timeline (next steps until the end).
3. need to write the 1 year report.4
4. Defining the prototype, reviewing objectives, hypotesis, variables, methodology (this is a subject that is coming back in papaer revies and doc consortioums that I was declined...too 'high level' is the general comment).
5. Updates about my qualitative analysis on the web 2 drag and drop experiment.
Remember: Metrics by Fixation Duration, Total Fixation duration, fixation count, fix count including zeros
Very interested on what happens when one comes across a novel form of interface, specially Post-WIMP with NUI modes of interaction
Very interested on the experience of what happens when one is offered an interaction with multiple inputs (gesture, eye gaze, voice) and with novel GUI solutions.
Present more graphics to justify my point!
Present more percentages within the problem. How many r interested? Any info online?
It's happening. But I should run a questionnaire if people r interested on engaging on Post-WIMP NUI
Studies about linguistics/natural gestures/gaze commands for the new interface
That was not done before. What am I improving? What novelty am I bringing?
Am I developing a technique/protocol? Recommendations for design?
Add a slide with completed work so far with checks (eg bibliographic review)
Check success/failure rate bibliography
Tool to capture websites: captures AJAX-style/SSL/Data mining (web mining)
RECOGNIZE is better then RECALL
how to analyze user's memory? What theory should I use? Within Cogn Psy?
How to analyze behaviour? Gestures language, grammatics, shall I bring back semiotics, linguistics
Define properly my variables
Correct my doc consortium doc
Declare general and specific obj
FRAMEWORK FOR EVALUATION
What is the Scenario?
Wednesday, 22 June 2011
1st City University Researcher Symposium: My Poster
Researchers’ Symposium
Opportunities for PhD researchers
PhD researchers are invited to participate in the Researcher’s Symposium on Wednesday 22nd June 2011. This will be at a prestigious event with a high ranking keynote speaker.
Why present a paper or a poster?
This is an ideal opportunity for students who have undertaken some data collection and analysis and would value some feedback as well as an opportunity to present in a conference environment. Presenters will gain experience in the process of presenting at a conference in a supportive environment, and will have the opportunity to receive feedback from the research community here at City University London.
Get the full programme of the day here
My poster:
Suggestion for topics for an improved version of my methodology of research:
Aim
Background
Data
Results
Discussion
Conclusion
Some thoughts about my research that emerged during the sessions:
It's well known that eyes seek eyes [ ]. We react when in others presence. Now comps seek eyes. Why shouldn't comps seek our eyes and react to our gaze and presence? It's only natural.
1. DISSECT GUI WITHIN POST-WIMP (QDA - Qualitative Data Analysis)
<tested with web 2 drag and drop widgets within PWP>
Assumption: Regular desktop GUI does not convey/signal properly possible interactions with NUI modes of interaction.
2. CREATE A PROTOCOL FOR UT: OBSERVATION / TASK
3. DISSECT THE GUI WITHIN THE PROTOTYPE (QDA)
4. CREATE A PROTOTYPE (POST-WIMP layer / NUI MODE OF INTERACTION layer)
5. METHODOLOGIES AND NEW THEORY for Post- WIMP embedded in NUI modes of interaction.
Check assumptions about traditional modes of interaction.
Visual cue that signals NUI?
You take for granted double click - it's a convention, it was learned - the UI itself does not provide a Perceptible Affordance for this.
Could adding eye gaze in combination with traditional methods for input slower down or speed up the interaction with a Post-WIMP UI?
When you gaze what is a double click and what is a simple selection?
How to right click?
Eye gaze for selections does not require learning
Check track pad learning
8 stylus pen
Opportunities for PhD researchers
PhD researchers are invited to participate in the Researcher’s Symposium on Wednesday 22nd June 2011. This will be at a prestigious event with a high ranking keynote speaker.
Why present a paper or a poster?
This is an ideal opportunity for students who have undertaken some data collection and analysis and would value some feedback as well as an opportunity to present in a conference environment. Presenters will gain experience in the process of presenting at a conference in a supportive environment, and will have the opportunity to receive feedback from the research community here at City University London.
Get the full programme of the day here
My poster:
Suggestion for topics for an improved version of my methodology of research:
Aim
Background
Data
Results
Discussion
Conclusion
Some thoughts about my research that emerged during the sessions:
It's well known that eyes seek eyes [ ]. We react when in others presence. Now comps seek eyes. Why shouldn't comps seek our eyes and react to our gaze and presence? It's only natural.
1. DISSECT GUI WITHIN POST-WIMP (QDA - Qualitative Data Analysis)
<tested with web 2 drag and drop widgets within PWP>
Assumption: Regular desktop GUI does not convey/signal properly possible interactions with NUI modes of interaction.
2. CREATE A PROTOCOL FOR UT: OBSERVATION / TASK
3. DISSECT THE GUI WITHIN THE PROTOTYPE (QDA)
4. CREATE A PROTOTYPE (POST-WIMP layer / NUI MODE OF INTERACTION layer)
5. METHODOLOGIES AND NEW THEORY for Post- WIMP embedded in NUI modes of interaction.
Check assumptions about traditional modes of interaction.
Visual cue that signals NUI?
You take for granted double click - it's a convention, it was learned - the UI itself does not provide a Perceptible Affordance for this.
Could adding eye gaze in combination with traditional methods for input slower down or speed up the interaction with a Post-WIMP UI?
When you gaze what is a double click and what is a simple selection?
How to right click?
Eye gaze for selections does not require learning
Check track pad learning
8 stylus pen
Markers:
Doctoral Consortium,
Eye Gaze,
eye tracking,
phd
Subscribe to:
Posts (Atom)

