This research used user’s relevance feedback to construct
more useful search query terms. The major weakness of the
general relevance feedback is the frequent user participation.
This may not be appropriate for the user who wants to
retrieve information immediately. Meanwhile, in the case of
relevance feedback based on a retrieval system, the retrieval
effectiveness significantly depends on the results of the
initial retrieval. Therefore, this research proposes a term
cluster query expansion model based on classification
information of retrieved documents, as a natural language
retrieval system that can accept infrequent user input to
judge the relevance of the results of the retrieval. This model
also leads user to participate in selecting term cluster from
the term clusters provided by the retrieval system.