This thesis presents the results from the design and implementation of condensing algorithm for Korean text. Condensing, here, involves a form of semantic processing and results in rough summary. The study requires various properties that general texts entail. Each sentence is characterized by constituents that make up the sentence, connective suffixes, and embedded clauses. Semantic and pragmatic aspects of Korean sentences along with syntactic information contribute to the decision process of condensing.
Korean text Condensing System(KCONS) consists of morphological analysis, co-word processing, sentence structure generation, and real condensing. According to the condensing degree, there are three kinds of condensing - low level, general level, and high level condensing.
The evaluation criteria are prepared to assess the performance at the system. Test results on sample texts are also presented.
