In this paper, a methodology for automatic coding is suggested which is fairly general. An application in the area of business applications has served as the inspiration for the approach is described.
Characteristic for this approach is that each code (say for a business activity) is characterized by one or more combinations of code words, or ð¶âwords. These combinations of ð¶âwords can be seen as definitions of the various codes that are used. It is assumed that the order of the ð¶âwords is irrelevant to describe a code. Because it is unlikely that people will use exactly those ð¶âwords when describing a business activity, synonyms, hyponyms and hyperonyms for the ð¶âwords are needed as well. These words connect the codes used in the classification with the descriptions provided by the respondents. A semantic network is used to link the descriptions to the codes. The paper describes the approach, and also how to produce such a semantic network, and which software tools are handy to support this activity. This paper is an updated version of [7]. As the original version of this paper was written in an older version of Word, it was decided to convert to LATEX for the new version, for which all figures were redrawn and in some cases replaced by improved versions.
