Free Model of Sentence Classifier for Automatic Extraction of Topic Sentences

M.L. Khodra (School of Electrical Engineering and Informatics, Bandung Institute of Technology, Indonesia)
D.H. Widyantoro (School of Electrical Engineering and Informatics, Bandung Institute of Technology, Indonesia)
E.A. Aziz (Faculty of Language and Arts Education, Indonesia University of Education, Indonesia)
B.R. Trilaksono (School of Electrical Engineering and Informatics, Bandung Institute of Technology, Indonesia)

Journal of ICT Research and Applications
Vol. 5 No. 1 (2011)

This  research  employs  free  model  that  uses  only  sentential  features without paragraph context  to extract topic sentences of a paragraph. For finding optimal  combination  of  features,  corpus-based  classification  is  used  for constructing a sentence classifier  as the model.  The sentence classifier is trained by  using Support Vector Machine  (SVM).  The experiment shows that position and meta-discourse features are more important  than syntactic features  to extract topic  sentence,  and  the  best  performer  (80.68%)  is  SVM  classifier  with  all features.