2007 Volume 24 Pages 139-148
We conducted a fundamental study for automatically abstracting citizens' interest toward infrastructure project by using interrogative sentences of the open-ended text data from public involvement activities. We also focused on features of the text, particles, adverbs and expression of sentence end. A Support Vector Machine-based automatic abstracting system was developed and applied to interrogative sentences from citizens in various PI methodologies. It is found that the interest by citizens were automatically classified into the following four categories named “Asking”, “Claiming”, “Dissatisfaction” and “Objection”.