Project Journal                                                                                                                                            Back to Home
Pre-Week 1

Week 1

Week 2

Week 3

Week 5

Week 6

Week 7

Week 8

Week 9

Week 10

Week 4 (June 10 - 17)

Read project related articles: "The PNA Project," "Interaction Design for Recommender Systems," and "The Role fo Transparency in Recommender Systems."
Read ObiComp papers by Weiser for Ph.D. students prelim discussion group.
Confirmed dual household questionnaire completion.
Reviewed Intro to Stats book.

This was a big reading week for me.  I started off with reading the paper "The PNA Project," (Personal Nutrition Assistant Project) which was a joint project between the the Computer Science department at the University of Scranton and the Nutrition & Dietics Department at Marywood University.  (I found the paper through the ACM and can't provide a link to it). 

The purpose of The PNA Project is to create an on-line tool for nutritional analysis of entered daily diets.  The device is similar to ours in that it takes into account personal health conditions (e.g. diabetes), and it utilizes the USDA Nutrient Database.  It differs from ours in that 1) it is more data entry intensive.  As is the nature of ubiquitous computing devices, our goal is to minimize the amount of data entry the user must do (low-fidelity sensing).  2) the PNA project provides a nutritional analysis of the inputted data.  We provide alternative suggestions based on nutritional analysis.  Maybe we should consider providing nutritional analysis as output as well.

I read the papers "The Role of Transparency in Recommender Systems" and "Interaction Design for Recommender Systems," both by Rashmi Sinha and Kirsten Swearington, to get a bigger picture of some user interface design issues we should keep in mind for our device.

Transparency is the process of conveying to the user the inner logic of how the system determined a recommendation. In "The Role of Transparency in Recommender Systems," Sinha and Swearington hypothesized that the less a user understands why a recommendation is made, the less likely he/she will trust the system suggestions. They went on to prove their hypothesis in a study involving music recommender systems.

Transparency is definitely an issue we should take into consideration in designing of our device. We want users to understand why we made certain suggestions regarding their nutrition and consumption, and thus trust our system. I personally would want to know why a certain type of food is better than the one I purchased. For instance, providing the nutritional analysis of both the purchased foods and suggested foods could be a major contributor to the transparency of the system.

In "Interaction Design for Recommender Systems," Sinha and Swearington discussed several findings regarding the design of effective interaction. One finding that I felt is relevant to our project is the observation that users don't mind providing more input in return for more effective output. Though we do not want to bog down our users with input, we shouldn't be afraid to ask for additional information (such as personal information) if that would improve our suggestions. The authors found that users dislike bad recommendations more than they dislike providing a little more input (in their study, input was in the form of ratings).

I also read this week "The Computer for the 21st Century," by Mark Weister as part of the Ph.D. students' reading list discussion group. It's one of the foundation papers of ubiquitous computing (ubicomp), was my first introduction to the field, and provided me with a good overview of how it started and where the field could be going.

Important points I found: "The most profound tehnologies are those that disappear. They weave themselves into the fabric of everyday life until they are indistinguishable from it." These opening sentences sum up the vision of ubicomp. Personal computers alone cannot be the future of computing; they alone won't make computing an integral, comfortable part of every day life. The future involves ubicomp, with invisibility being a major component of computing. Multi-media and virtual reality (VR) can't define ubicom because multi-media machines make the computer the center of attention, and VR focuses on leaving the world rather than on invisibly working within the world. These are important distinctions because the goal of ubicomp isn't to put computers everywhere, overwhelm users with technological options, and to make computers the center of attention. The goal is to push computers into the background and to make people, not computers, the center of attention. I found this be very important information because as a novice to the field, I intitially felt hesitation about the idea of making computing even more ubiquitous. I know many people who feel they have more technology than they can handle already. If the goals and future of ubicomp are as Weiser described, than more people, including myself, would be open to it.

I also read Weiser's "Some Computer Science Issues in Ubiquitous Computing" in which he described ubicomp largely based on prototypes developed at Xerox PARC (the tab, pad, and board). Weiser discussed the role of invisibility in ubicomp, details of the PARC prototypes, hardware requirements and network needs of ubicomp, and the issue of preserving privacy.

This week I also confirmed the possibility of adding to our long-term survey a dual household consisting of a parent and child. The household agreed to be in our candidate list and to complete two questionnaires if selected. I also reviewed a statistics book which is in the lab collection. The book was kind of dry, but it was good starting to think about how we can organize the data.

< Previous Week          Next Week >