Theme 4

My article this week is “Qualitative evaluation of automatic assignment of keywords to images” by Chih-Fong Tsai , Ken McGarry and  John Tait is once again a image & video inspired article, but not as technical as the ones I usually choose. It comes from the journal “INFORMATION PROCESSING & MANAGEMENT” which just has an impact factor of 1,119 which is the lowest IF I have had so far. This article is about how humans assign keywords to images in a different way than computers. This includes a qualitative method where five people assign keywords to images and then this is compared to what the computers have annotated to this image. This is a part of the research field IR, information retrieval which is becoming a bigger and bigger research-field today.

I think this is an important research because it shows that in this case the computers may today have a problem with being better that the computers. To make a qualitative research on this is good because it can show more and deeper information than if it would have been a quantitative research. It is the process that is important and it is a flexible research. The problem with this research and qualitative research in the whole may be that it is a very small sample of 5 people. They may show some important things, but it is still few people. It may be subjective as well. It is hard to say whether there are other things left out in this research since they use some models that is limiting. The evaluations model which they base their qualitative method on is called Type I and Type II evaluation models and this is models for evaluations and data collections from human judgments. I think this is great models for evaluating human judgments where it works as guidelines for how to analyze behavior. The models are open and this can be a benefit where it is used in a way which fits right in a research but I also think there are big risks with these kinds of frameworks with smaller space for innovation.

The article  “Comics, Robots, Fashion and Programming: outlining the concept of actDresses” by Fernaeus, Y. & Jacobsson, M. (2009) is a interesting article in a pretty special subject I think. I like the approach of learning things from two totally different areas like sign systems from comics and from how people are clothing their robots, like Roomba and Pleo. To link these parts to physical programming looks for me as a nice way to invent new ways to solve the old problems. This is a beautiful way to be innovative. In this example when they use clothing to control different kinds of robot they have found a innovative way to solve a known problem. I am interested in this kind of findings and how you can come up with them. I think this is the beauty of qualitative methods where you can explore things in a deeper way. Quantitative methods are still more of repetitive and only confirming way to work where qualitative methods are the innovative way.

My question is if these finding and this progress they did in actDresses could have been done through a quantitative method or if this type of innovative new finding are exclusive for qualitative methods?