I am at the IBM: New Directions in User Computing conference. It is the annual user interface summer conference held at IBM every summer. During Dan Russells’ and Steve Cousins introduction, we learnt that this is the 13th such conference. This year, the focus is on mobile computing.
Its about 100 or more people. I saw a lot of Bay area HCI researchers (the same ones that you run into at CHI conference). Unsurprisingly, there are quite a few IBM researchers and interns. Also, unsurprisingly, there are very few practitioners. This is mostly the CHI crowd – i.e., mostly researchers.
I was there for just the morning. Really enjoyed a talk on Social Mobile Computing by Ian Smith from Intel Research Laboratory in Seattle.
Here are my notes for the talk.
-No one has figured how to get people to buy apps for the mobile phone. Ian asked audience how many people had bought apps for mobile computing. Very few had.
-Whoever figures this out will be rich, retire to Jamaica…
-Ian thinks social mobile apps will be first apps for the mobile phone that people will buy.
–KRUMM: place studies
–PARC Nextel studies of PTT
–Nokia: DeDe, Scent, Digidress
-For his own research program broad goals are to develop algorithms, tools around mobile computing.
-One basic premise behind all these types of apps is “Location Assumption” or the fact that you need to know (at least approximately) what the users location is.
Question is how to do that. GPS is one method. However, it does not work inside buildings. Additionally, it does not have a Z dimension. So, we will not know what floor we are on.
-Describes how to use triangulation: radio beacons such as 802.11 access points, GSM cell phone towers, and fixed Bluetooth devices that already exist in large numbers around us in the environment to identify location. Ian showed a map of all San Jose locations he drove to yesterday: gps tracks and cellular signal strength. Overall, you can make a reasonably accurate assumption about the location.
New Application ideas from Intel Research Seattle
– (Jeff Hightower, A Varshavsky)
–Phones have a call log. why not have a place log.
–Algorithmic challenges are numerous: being specific about location is difficult.
–Ontological or personal naming problems are many (what does place mean to you) 46.133 North might not mean much to me, but the Chinese restaurant on El Camino might be very meaningful.
–If you could have some sort of naming, tagging, you could have many useful outcomes. For example, you could personalize yahoo directions (e.g., get personalized directions to the closest Chinese restaurant), and to the bakery on the way back.
Albany: Recommendations through location histories (M. Chen/J Froelich)
Basic premise: If there are types of places you always go to, then system can recommend other, similar places to you based on your current location. For example, there might be a particular type of bar you frequent when in your home location. You are in Seattle right now. You might like a recommendation about such a bar near your Seattle location.
One can also use collaborative filtering (other, similar users liked these places) for recommendations.
Potential problems: Data churn can lead to a lot of problems
-for example, if restaurant location changes or restaurant closes down.
-or if cellular tower changes, and you can no longer triangulate location.
Stimulating the episodic memory can prevent cognitive decline
-make it easy to capture a memory on a mobile phone
–events in between trip to Europe and a daily blog
Playback tool that uses media clues to get you to relive the experiences.
(Not sure I buy into this idea. From my understanding stimulating episodic memory is not the key to prevent cognitive decline. This seems somewhat of a stretch, a technology in need of an application).
Playback Avec Vous
Iain Smith, Patrick Bayudish, D Tan., K Inkpen
-Many cell phone conversations start with “Where are you”
-An application could let people know where you are.
-Privacy sensitive, map based location application for finding someone’s location, or disclosing your own.
-Helpful for family coordination, and rendezvous planning.
-Built on ideas derived from Dodgeball.
-Privacy issues: You can disclose your own location, but not track others
-Can also send someone a location in the future. lets meet at point X (and send a dot to that point).
Health Apps: Houston (Consolvo, Everitt)
Sharing stepcounts with your social network
GYm partner effect: if you tell your friend you will meet them at gym, more likely to go
you can share your pedometer count with friends. This will make you more motivated.
We can make a guess at activity based on location: If I in Starbucks, chances are – I am having coffee. In a Dry Cleaning place, picking up or dropping off laundry. By finding location, you can get some idea about someone’s current activity.
Example application might be to for a health care worker to track well being of elder person they are caring for. If you locate them in parking lot of hospital, it might be need to be further looked into.