MAS965: Social Visualization
Prof. Judith Donath
Assignment 8 - The Face
Nov 16, 2004

Francis Lam

Oscar Wilde's novel The Picture of Dorian Gray, is based on the notion that as people age, their face reflects the type of life they have lived. In this story, Dorian Gray possesses a portrait that ages instead - Dorian remains ever youthful, while the portrait, which he keeps hidden away, displays his cruelty and excess. Sketch an interface for a system in which user's history was etched in faces. Useful scenarios to ground your work might be reputation systems such as ebay or conversation systems such as slashdot. What information would you depict with this face? How would it "age"?

I try to incorporate human faces in reputation systems such as eBay. My idea can be structured in the following points:

Input: users can choose a face representing their identity on the system

Interact: users can give positive, neutral or negative feedback to another user by putting different stamps on their face

Output: the expression on a user's face is illustrated according to the feedbacks

Individual identity
First, the system provides a series of generic cartoon style faces for selection. User can have limited customizations, such as the silhouette and the color of hair, to the face they have selected to represent the kind of image they would like others to perceive them as.

Social identity
When you are giving feedback to another user, you can also choose a graphic to stamp on the target's face. The graphic should carry a positive or negative meaning in our common understandings and be appropriate to be put on a face. In the sketch, I use a kiss-mark for a positive feedback and a cross-mark for a negative feedback. I understand that putting graphics on people's face would raise many problems due to different cultural norms, but I hope to use cartoon- and cheerful-like graphics to soften some of these problems.

Not only is a point-scale system displayed in a user's profile like the current system, it also displays the face selected by the user with the facial expression according to the feedbacks given by others. The expressions in the system could be prototypical, happiness and sadness, as the parameters are simply positive and negative feedback.

When clicking into the detailed view of a user profile, you can see the history of positive or negative feedbacks left on his face with the opacity of the mark showing how recent it is.

What is the difference between the "emotion view" of the face and the "evolution view"? What are the implications of each standpoint for the use of the face as input? As output?

The notion of emotion view is that the human face expresses the underlying emotion of the human. It supports a direct link between what appears on someone's face and how they feel inside, so even if they try to mask their emotion, there should still be detectable clues on their face. The belief of evolution view is that facial expressions forecast people's future actions, the expression on people's face can tell outside observers what they will do next. The main difference between these two views is from whose point of view they explain the facial expression. Emotion view focuses on the person as the expresser while evolution view sheds light into the observer who perceives the expressions which implies that expressions can be signals social intent.

The emotion view holds that facial expressions are universal, so it is plausible to imply that a system which takes human faces as input with proper technology and context-awareness, is able to map the faces to a pool of prototypical facial expressions which represent certain emotions. And as output, prototypical faces can be used as indicators of users'emotional status in a reputation or a conversation system.

The evolution view supports that faces and expressions take a very important role in catalyzing social interactions by forecasting each participant's future move. It has implications for real-time human-computer interaction, both for human face recognition and computer agent generation. As output, the agent's face should provide signals of the system's future action and also provide facial feedback that the user's expressions are being received. On the other hand, the user's face recognized by the system should also supply information about the user's emotional status to anticipate what he/she will do next through a learning process about each user.