IA Summit 2004 Emotions In Information Seeking
These are my notes from Jim’s presentation on Feb. 29.
Discussions of emotions have been around for a while so we aren’t talking about a paradigm shift. For example Krug’s “Don’t Make Me Think” is really don’t make me frustrated.
Ellis behavioral model for information seeking, Marchionini’s model. More recent Foster (2004) criticizes for not accurately reflecting user behavior but Ellis’s wasn’t meant to be linear so I don’t really agree w/ those criticisms.
Carol Kuhlthau--perception of complexity rather than the actual complexity that causes anxiety.
One would assume that confidence increases during a search, but Kuhltau found opposite particularly in stage 3 and 4.
an information search process (ISP) is limited to a defined set of users (referenced George's persona presentation)
(showed matrix for job site) moving between 3 and 4 we found uncertainty like Kuhlthau did. The monitoring stage was important.
How do we measure emotions? with time and money we can do like Richard Hazlett who attached electrodes to measure face muscle movement (See CHI 2003 Measurement of User Frustration: A Biologic Approach). Most of us can't do that, but I'm advocating more traditional role playing interviews, observations, customer feedback.
- needs - problem, gap
- internal - driven by prior knowledge and experience. unique to the user's profile
- External - context for information seeking
information search experience model
(Shows summary for job site highlighting the planes.) Taylor's problems map to my needs column. So how can this be helpful share a common vision for the development team? Often we (wrongly) talk about what it is to be developed rather than how we're going to develop it. Structure user research (referenced george's persona presentation)
uncertainty and complexity as heuristics in search design
Audience member: developers say that the machine will disambiguate the users request I say its the user. I say its the user who determines the relevance.
Jim: right we know there are at least 4 different types of relevence. when developers talk about relevence they are really talking about algorithmic relevence.
Audience member: have you compared with information encountering behavior?
Jim: yeah that's more what foster was talking about. I still think a linear process is more applicable to the design process. Foster is great for research and it asks more questions.
Audience member: Marcia Bates' berry-picking?
Jim: at some point in the design we have to define our groups and typical behavior but we can capture that in the description of the behavior. Define the emotional event that causes that pivot.