The value of geographic Wikis.

Abstract

This thesis responds to the dual rising trends of geographic content and open content, where the core value of an information system is derived from the work of users. We define the essential properties of an emerging technology, the geographic wiki or geowiki, as well as two variations we invented: the computational geowiki, where user wiki input feeds an algorithm, and the personalized geowiki, where the system provides a personalized interpretation. We focus on two systems to develop these ideas. First, Cyclopath, a research geowiki we founded, serves the bicycle navigation needs of cyclists. We also present analysis in the context of Wikipedia, the well-known and highly successful wiki encyclopedia, using its size and maturity to draw lessons for smaller, younger systems which are far more numerous but hope to grow. We ask three questions with respect to this new technology. First, can it be built? Yes. This thesis describes the design and implementation of Cyclopath, which has grown to be a production system with thousands of users. Second, is it useful? Yes. We identified a representative geographic community, bicyclists, and they both tell us that the information in the Cyclopath geowiki is useful and show us by using the system in great numbers. We also present new ways to measure value in wikis, introducing new techniques for doing so from the perspective of information consumers. In particular, user work in Cyclopath has shortened the average route by 1 km. Also, we present techniques for obtaining more contributions (familiarity matters -- sometimes -- and users do work beyond what they are asked to) and algorithms for increasing the value of geowiki content by personalizing it, showing that traditional rating prediction algorithms (collaborative filtering) are not effective but simple algorithms based on clustering are. Finally, who cares? Many people. There are numerous communities with great interest in geographic information but limited, incomplete, or awkward access because the relevant knowledge is distributed among members of the community and otherwise unavailable. As our results demonstrate, geowikis are an effective way of gathering and disseminating geographic information, more so than previous techniques. Thus, this research has broad value.