印尼文文本摘要的句子提取和还原-sentence extraction and restoration of indonesian text abstract.docxVIP

CONTENTSCHAPTER 1 INTRODUCTION1Motivation for this Thesis2Project Overview3Thesis Outline3CHAPTER 2 TEXT SUMMARIZATION REVIEW5Indonesian Language5Information Retrieval7Na?ve Bayes Model8Na?ve Bayes Theory and Example9Bayes Na?ve for Sentence Extraction 11Hidden Markov Model (HMM) 12Decoding using Viterbi Algorithm 14Hidden Markov Model for Sentence Reduction 16Related Work 18CHAPTER 3 NA?VE BAYES FOR SENTENCE EXTRACTION21Probabilistic Model 21Na?ve Bayes Application 22Document Representation 22Text Features 23Semantic Feature 25The Implementation 27Sentence Extraction Process 27System Architecture 28CHAPTER 4 HIDDEN MARKOV MODEL FOR SENTENCE REDUCTION29Probabilistic Model 29HMM Topology 31The Implementation 32Preprocessing 32Sentence Reduction Decoding 33Bigram Smoothing 34Grammar Checking 35System Architecture 36CHAPTER 5 EXPERIMENT AND EVALUATION RESULT 37Training Corpus 37Evaluation of Sentence Extraction. 38Performance of Single Feature 39Extraction Evaluation 41Evaluation of Semantic Feature 41Evaluation of Sentence Reduction 42Intelligibility Evaluation 42Metric Evaluation 43CHAPTER 6 CONCLUSION 46Conclusion 46Future Work 47Acknowledgements. 48References. 49APPENDIX A: STOP WORDS 53APPENDIX B: EMPHASIZE WORDS FOR NEWS 57APPENDIX C: TAGSET USED FOR BAHASA INDONESIA 58APPENDIX D: CUE PHRASE 59APPENDIX E: GRAPHICAL USER INTERFACE 60APPENDIX F: SUMMARIZATION EXAMPLE 63CHAPTER 1 INTRODUCTIONIn the past, information was something very rare and only accessible to those belonging to a truly restricts elite. In the 1990, since the beginning information age and especially when the quick expansion of Word Wide Web (WWW), has produced a huge amount of many kinds of online document such as: text, image, audio, and video. Now, we live in a wonderful era, with information readily available at our fingertips. With access to the Internet and the entry of a few keywords on a search engine, we can be instantly rewarded with the information we are hunting. The spreading of Inform