The Final Week!

This week we have been working very hard on preparing our presentation. We participated in a group conference call on Sunday, December 9 to discuss our ideas and the work we've completed for the presentation. Edith has updated the task flow for the Voice Interaction. She has also created a PowerPoint presentation draft. Lisa is working on the prototype redesign and creating presentation examples. Karen has been working on presentation graphics. We will be meeting an hour before class tomorrow to practice and synchronize our presentation.

Lisa's list of tasks for herself:
- demo (Tues)
- slide draft (today)
- design document (tomorrow; includes Edith's workflow + bubbles,
grammar description, examples)
- handout (Wed)
- CD (Wed)

Conducted Participatory Design Session 2

We held our second Participatory Design session on Thursday, November 29. It went as well as the first session! The participants had fun testing the Coffee Sleuth prototype and the Google Mobile and Tell Me services. They gave us some valuable suggestions that we can incorporate into our prototype.

Leading up to the session, the group collaborated a great deal via Gmail and in person to plan it. Edith was the primary facilitator while Lisa assisted. Our activities leading up to the session included:

• Group planning
• Lisa designed the prototype.
• Lisa and Edith met to further plan the facilitation of the session just prior to it.
  

Each of us took notes.

Next steps:

• Prototype design revision -- Lisa
• Re-design task analysis -- Edith
• Summarize the second session -- Karen
  
• Design a new logo and graphics for the presentation -- Karen
• Propose presentation ideas and design the presentation -- Group

Since the session, Karen has summarized the notes and designed a new logo based on an inspiration from the session. Edith has been working on the task analysis. Lisa has been working on the prototype redesign. In addition, we have shared our research papers with each other.

Thanksgiving - and design work

We're all writing and sharing research information. Lisa worked on the group design very heavily.

This is an abstract for our presentation. We still need to put it together... but the design work is central and needed to be done early.

We will describe a mixed modal, speech and text message conversation interface that maintains contextual understanding between exchanges. The domain is restricted to finding coffee shops and the interaction is modeled as an information-seeking dialogue. We focus on designing a generic service that works broadly across all telephone networks and cellphone models. Our goal is to improve both the efficiency of search and relevance of search results - as well as to improve the experience of search on cellphones.

I don't want to give away everything... but here is a synopsis:

Considering our initial study, design goals for CoffeeSleuth are the following:

Innovative messaging behavior (dialogue interaction)
Flexible interaction (speech and SMS)
Social engagement (social networks and user-contributed data)

One caveat: we want no reliance on custom, user-installed software; we will use Speech, SMS plus MMS, WAP Push or WAP browser

CoffeeSleuth extends traditional Voice and SMS services in the following ways:

1) Flexible dialogue interaction:
- The incorporation of a "route" plan
- Semantic accommodation for search filters and pronominal reference across messages
- Ability for users to examine information spatially or in other ways before selecting a particular location ("map all")
- Additional contexts (social, dialogue, situational)
- Integration of voice and SMS into a single threaded interaction

2) Social engagement:
- Add social data. Allow users to “digg”, add photos, add text comments and tag.
- Apply search filters on the basis of social data
- Create a network of friends (using Facebook webapp)
- Save location to favorites

A key design decision was to “not conquer the world.” We offer a very focused application service and leverage other existing services in deployment. CoffeeSleuth is currently deployed as a simple demo on tellme.com and makes use of freely available APIs such as Yahoo Local Search, FaceBook, etc. Our value-added is our flexible interaction and more relevant results due largely to user-submitted content. The more user use the system, the better their experience becomes.

Is this technically do-able? Yes! We're leveraging existing technologies and infrastructure. Having some with a computational dialogue background ensures we're grounded in reality.

We still need to do our second participatory design session. We're looking at the 29th and potentially a modified WOZ scenario.

Research!

All of us are in heavy research mode and have been sharing research information. Karen has decided to focus on speech systems for her research paper. Edith is focusing on context-aware systems. Lisa is working on conversational systems. She suggested that all play with both Google (SMS and GOOG-411) and Tellme.com systems. Here are some speech transcripts and un-polished notes from her experience.

Google Mobile SMS - 466453
http://www.google.com/intl/en_us/mobile/sms/#

- Type the name of the business or service followed by zip or city and state
- Alternatively, "SET LOCATION city/state" and all queries that follow will be at this saved location
- Directions FROM city/state TO city/state

If you send a blank message to Google Mobile, you get back a help message that tells you about all of the searches you can make. Everything is abbreviated. Here is the text:

Text to 466453 to get results, e.g,;
Local: blockbuster sf
(W)eather: w nyc
(M)ovies: m 94114
flights: ua 14
Sports: ny jets
Stocks: goog
More help: tips

When I responded with local I got:
To get business listings, enter what you want to find and include a city and state, or zip (ex: pizza 10013, Blockbuster boston ma)

Entering in an address, I got back one search results. I would have liked to have been able to request more.

1-800-Goog-411
State LOCATION and BUSINESS TYPE
Things you can always say: "Connect", "text message", "map it"

----------------
Goog-411 transcript

Prompt: What city and state
User: Clarksburg, Maryland
Prompt: What business name or category?
User: coffee
Prompt: Searching
Prompt:
Top eight results
Number 1 Myorga Coffee on Stringtown Road
To select number one, you cna press one or say number one
Number 2 Starbucks on Frederick Road, Germantown
Number 3 Starbucks on Frederick Road, Germantown
Number 4 Starbucks Coffee on Germantown Road, Germantown
To start a new search, say start over anytime
Number 5 Starbucks on Frederick Road, Germantown
Number 6 Dunkin Donuts on Wisteria Drive, Germantown
Number 7 Mayorga Coffee on Wisteria Drive, Germantown
Number 8, Aqui Brazilian Coffee on Wisteria Drive, Germantown
Top of the list
User:
Number 1 -
Prompt: Number 8, Aqui Brazilian Coffee on Wisteria Drive, Germantown
Prompt: I'll connect you or say details or go back
User: Text message
Prompt: Hold on, Sending text message
Prompt: Hang on and I'll connect you, or say go back.

Basic dialogue flow:
CITY/STATE
BUSINESS NAME or CATEGORY
CHOICE

Voice - very nice quality. SR very good. System very usable until the end when it expects you want to be connected automatically. Eeek! All alternatives are not provided via the final choice menu. System choices easy to use - top eight listings and street number. Use of keyboard input for number - mixed modal interaction. Some small choice-based interaction on webpage output.

SMS output:
Contact info -> WAP link to a map -> WAP link to driving directions (from default location)

SMS service has a very simple dialogue functionality of "remembering" a default location.

-------------
Tellme transcript
Prompt: Main menu
Prompt: business search, ring tones, stock quotes, news center, sports, entertainment, travel, and popular shortcuts
User: business search
Prompt: hmmm. I didn't get that. Please say one of our categories or say main menu
Prompt: business search
Prompt: I'm sorry, I still didn't get that. To hear all the categories we have, say main menu
User: business search
Prompt: I'm sorry. I'm having trouble understanding. You can also punch the first three letters of one of our categories on the telephone keypad, for instance -- okay, business search then.
User:
Prompt: city and state please
User: clarksburg, maryland
Prompt: okay. what business, or type of business are you looking for?
User: coffee
Prompt: To narrow your search, you can say by intersection or you can say I don't know
Prompt: Alright. When you hear the one you want, just say the number.
Prompt:
1) Mayorga coffee
2) Starbucks, in multiple cities - (There are two. Do you want the __ or the __)
3) Music Cafe, in Damascus
Prompt: You can say repeat, tell me more choices, or new search
Prompt: Sorry I didn't hear you.
User: Tell me more choices.
Prompt: More choices
Prompt:
4) Dunkin Donuts in multiple cities
5) HUUI Brazilian Coffee, in Germantown
User: 5
Prompt: Okay, HUUI Brazilian Coffee at the 12615 Wisteria Drive Germantown Square
Prompt: The number is 301-540-3395. I'll text you that info now.
Prompt: You can say share this listing, you can also say repeat the info or start over.

Voice - very nice quality main interlocutor. 'Choice reader' very mixed... Different voices mixed together. More prompts than Google, but very easy to use. I preferred this interaction since I wanted a text message response and not a connection. Speech recognition not as good, but text input alternative eliminates much frustration. Use of keyboard for input - mixed modal interaction. More efficient response than Google's when there are multiple alternatives of a chain.

Basic dialogue flow:
BUSINESS SEARCH
CITY/STATE
BUSINESS TYPE
CHOICE

Tell Me
1-800-555-TELL
"Business search"
NAME or TYPE
"Text me the info" - set to automatic
"Share this info" - with friend
"Go back" - wheew sound.

Tell Me Text 83556

"Coffee Clarksburg MD"
set default city. "d clarksburg md"
weather. "w clarksburg md"

After message receipt:

"m" to get more results
"map" to map
airport codes instead of city
wm - more results for weather

Downloadable - search using speech. feedback on screen.
Unable to test since none of our phones is supported.

Tasks Performed for week of Nov 2 - Nov 7

Karen has made us a very nice logo. We're posting it to the blog.

Edith and I gave Karen our notes from the participatory design session. She is working on the session summary.

My email shows that we have been working on researching ideas for CoffeeSleuth and sharing research information. We are honing in on a voice and SMS interface. I provided Edith and Karen with a little article on conversational interfaces: http://www.pbs.org/speak/ahead/technology/talking/

Group Activities for the week October 24 - Nov 1

We held our first Participatory Design session on Thursday, November 1. It went very well! The participants had fun with the scenarios and gave us valuable insights -- some that confirmed our initial design assumptions and some new ideas that we can incorporate into our prototype.

Leading up to the session, the group collaborated a great deal via Gmail to plan the session. Lisa and Edith swapped assignments as to who would plan for and facilitate the first session and who would facilitate the second session because Lisa had the creative energy around the design. Our activities leading up to the session include:

• Lisa developed a script for the first based on group discussions in class on October 24 that she submitted and on initial design ideas for the application.
• Lisa drafted two quick scenarios to guide participant interaction during the session.
• Lisa prepared the Human Research Adult Consent Form
• Karen and Edith provided screen captures of cell phone interfaces and sample applications that might influence the design of our application.
• Edith found a research article on a multi-modal mobile app that uses voice interaction along with text interfaces for navigation in strange city (to support the 2008 Olympics in Beijing). Lisa's response to the article helped to clarify our design approach as unimodal system with an underlying representation that could be extended to support multimodality.
• The group is continuing to research context-aware mobile services and speech interaction.
• Edith sent the formal meeting invitations to the participants for the two participatory design sessions (the second one is scheduled for November 14).
• Edith gathered the materials, including Starbucks coffee, for the session participants.

We recorded the session, which Lisa transcribed. Each of us took notes; Karen is going to summarize the notes.

Next steps:

• Summarize results of first PD session for use in Design: Karen
• Task Analysis & Design -- Edith
• Prototype Design -- Lisa
• Develop second session goals -- Edith

Tasks Performed for week of October 17 - 24

• In-person recruiting of participants for Participatory Design session
  
• Developed preliminary scripts for first participatory design session
• Design discussion of entry approaches to CoffeeSleuth Service

Next steps:

• Send out email invitations to PD participants confirming time and location -- Edith
  
• Finalize market "summary" description --Edith
  
• Draft preliminary task flows (not to share with users) -- Lisa (change in assignment)
  
• Draft and post scripts for first participatory design session -- Edith
  
• Market research on existing products and technology (SMS, etc.) -- Lisa
  
• Post notes on heuristic research -- All
  

Group Outline

1. Review heuristics research findings (Group)


2. Summarize "marketing" information –
   i.e., document assumptions about what the users want, priorities, etc. (Group)


3. Develop "high concept" statement to drive design (Edith)


4. Task Design (Edith)


5. Develop prototype


• Design (Lisa)


• Graphics (Karen)


• Rendering (Edith)


6. User recruitment for PD session


• Set dates for sessions (Group)


• Personal contact of user candidates (Lisa)


• Develop consent form (Lisa)


• Send out email invitations to confirm date, time, location, purpose, etc. (Eidth)


7. Prepare for first PD session


• Identify session goals (Group)


• Gather materials (Edith)


• Build props (Karen and Lisa)


• Develop session "script" of activities, questions (Group)


• Roles:


• Facilitator (Edith)


• Observer(s)


• Other?


8. Summarize issues from session (Karen)


9. Document task analysis findings (Edith)


10. Incorporate design changes based on PD feedback into prototype (Lisa)


11. Prepare for second PD session


• Identify session goals (Group)


• Gather materials (Edith)


• Develop session script (Group)


• Roles:


• Facilitator (Lisa)


• Observer(s)


• Other


12. Summarize issues from session (Karen)


13. Finalize prototype (Lisa)


14. Develop PowerPoint briefing (Edith) -- Graphics (Karen)

Group Research Assignments

Lisa Harper -- Cutting-edge mobile interfaces
Edith Hughes -- Service design for mobile interfaces; task analysis for mobile design
Karen Pascarella -- Psychology-driven IA for mobile interfaces

We have begun!

This is a group project for University of Baltimore IDIA 612 (Interface and Interaction Design).