Situated nature of similarity
This paper begins with a review of literature citing some works on the situated nature of similarity, an issue of some importance. The article moves on to deal with a partonomy approach that decomposes similarity into little bits to weight and sum up. While this may work in some cases, the evidence from ethnography points towards more complex kinds of similarity. The concept of boundary object (Harvey and Chrisman 1999) derives from a substantial literature of symbolical interactionists working on how divergent groups negotiate shared meaning. If we want to learn about similarity, we need to understand how each participant interprets the same stimulus. For me, I am reminded of some work done in the 1980s at University of Wisconsin. We found two departments of the Wisconsin government mapping the state using radically different measurement frameworks, yet a similar underlying objective. One was mapping 'wetlands' with polygons interpreted from airphotos. The other was assigning a code to a property parcel to account for 'wastelands'. By any text search, the two concepts are different, but operationally, they become quite similar. In reading the detailed manuals, wastelands were defined as 'bogs', 'marshes', and other terms that are also used to described wetlands. The real test was that most of the wetlands were classified as wastelands. The fact that some wastelands were not wetlands may be more a measure of political corruption than of formal semantics. The framework proposed in this paper provides a framework to deal with this situation, but it takes some finesse to interpret all the documents and trace what kind of similarity is justified for each participant.
