Paper Reading #25 - Finding Your Way in a Multi-dimensional Semantic Space with Luminoso

Comments

Shena Hoffman

Aaron Kirkes

Reference

Finding Your Way in a Multi-dimensional Semantic Space with Luminoso

Robert Speer, Catherine Havasi, Nichole Treadway, Henry Lieberman

IUI 2010 - Hong Kong

Summary

This paper focuses on an interface designed to aide researchers in visualizing a semantic space. A strong focus is placed on recognizing links between semantic data as well as identifying patterns in the data. The process relies on two earlier developed tools which create the semantic nodes of interest. A set of documents is input, with the option to mark certain documents as more important than others. Relationships are then determined between documents by comparing natural language patterns. A projection of the multidimensional data into 2D screen space is then performed -- yielding a circular like representation. The paper focuses on the action of “grabbing” which allows a user to select a particular semantic aspect and change the projections to gain more insight into connections.

Analysis

While this paper didn’t quite sell me on the “grabbing” feature (which it over emphasized), I think the overall interface and tools presented here would be quite useful when looking at large datasets. It would be interesting to take data from a user’s web browsing history and run it through this process. It would be particularly interesting to see what kind of links are derived by the algorithms as far as thoughts sparking other thoughts--- I think we might see some neat connections we wouldn’t otherwise think of. Overall this is an interesting tool and I’d be interested in experimenting with it.