Wednesday, April 20, 2011

Paper Reading #24 - Finding Your Way

Comments:
Comment 1
Comment 2

References:
Title: Finding Your Way in a Multi-dimensional Semantic Space with Luminoso
Authors: Robert Speer, Catherine Havasi, Nichole Treadway, and Henry Lieberman
Venue: IUI 2010, Feb. 7-10 2010

Summary:
In this paper, the authors describe a program called Luminoso that provides an interface for parsing text input and displaying it.

The Luminoso system displays the relations between text sets in N-dimensions. These dimensions are first created by finding the occurrences of words in each document, and then analyzing the meanings of each of the highest words. Then, the system gathers the words with similar meanings, and assigns a dimension to each gathering.

To examine the data, the program defines the interface above. The lines connect the words related to the current selection. The colors define how "hot" a particular relation is, from white being highest, down to red. The user can navigate by selecting a particular point and then rotating into the semantic dimension you want.

Discussion:
While I think it's important to be able to quickly navigate through text to find what you need for situations like surveys, I don't think that this is the best navigation method. The concept of n-dimensionality is confusing to start with, but when you add in the abstractness of the data you will be sorting through using these dimensions, and I would feel completely lost.

Also, this paper mentions a lot of sorting and data modification methods without defining them, so in many cases I was unable to understand the backbone behind how the system worked. I think they probably would have done better to lengthen the paper and add some short definitions for each term.

(Image courtesy of: )

3 comments:

  1. I had a hard time understanding what the paper was saying too. I thought it was poorly organized, and had to go back through it to really figure out what was going on. I don't think this system is that useful either.

    ReplyDelete
  2. I agree with you, this does not appear to be the best method for navigation through the document. I think this could be used in other applications though, like exploring topics in the paper.

    ReplyDelete
  3. I agree that this sounds overly complicated. Perhaps they could have done a study to verify the effectiveness of their system? It is somewhat ironic that they would publish a poorly organized paper on navigation.

    ReplyDelete