Aspects of sentence retrieval

Abstract

Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. I dedicate this thesis to my great-grandmother Mildred Graham who completed the sixth grade, taught her husband to read and put her son through law school, and to my great-grandmother Olive Peterson who completed a Masters in Mathematics, and put her daughter through the Sorbonne. My generation owes a debt to those who came before because they prepared the way for us, so that everything was easy, and we only had to work hard to succeed. We hope to return the favor by teaching our children to be strong and courageous, thereby making us worthy of the grace bestowed upon us. ACKNOWLEDGMENTSNothing worthwhile was ever achieved in isolation. I can not claim to have written this thesis without the significant influence of others. I would like to thank my advisor W. Bruce Croft. It was invaluable to have the benefit of his more than 30 years experience in the field of Information Retrieval standing behind his advising of me. I am most grateful that he allowed me to work with him for five years, and to be a part of the best information retrieval research group in the world. I would like to thank James Allan for serving on my committee. James sets a tone in the lab of friendly cooperation, which makes it possible to get more work done, and makes the work itself more enjoyable. I would like to thank David Jensen for the counsel he has given me in the time I have been at the University of Massachusetts, which was especially kind of him, considering that I am not a member of his lab. I would like to thank John Staudenmayer for serving as the outside member on my committee, to ensure the process is fair, even though he was already very busy. and who I admire for his courage, perseverance, and intelligence.I would like to thank my mother, Chris Jorgensen, who taught us to be strong and independent, and treated us as if we already were the intelligent, competent people she expected us to be. marization, novelty detection, and information provenance make use of a sentenceretrieval module as a preprocessing step. The performance of these systems is dependent on the quality of the sentence-retrieval module. Other tasks such as information extraction and machine translation operate on sentences, either using them as trainin...

A recursive TF-ISF Based Sentence Retrieval Method with Local Context

“…Another typically used method for sentence retrieval is the query likelihood method which is a language modeling approach to document retrieval. That method, invented by Ponte & Croft for document retrieval [11] was in a trivial manner adapted and often used for sentence retrieval [2]. The TF-ISF and the query likelihood method can be considered baseline methods for sentence retrieval.…”

“…The TF-ISF and the query likelihood method can be considered baseline methods for sentence retrieval. The TF-ISF method and the query likelihood method were used as the baseline methods for sentence retrieval in [1] and the query likelihood method was used as the baseline method for sentence retrieval in [2].…”

“…Tasks like novelty detection, question answering, summarization, opinion mining and information provenance make use of sentence retrieval [1], [2]. Sentence retrieval methods are usually simple adaptations of document retrieval methods where sentences are treated as documents [3]- [5].…”

“…Sentence retrieval methods are usually simple adaptations of document retrieval methods where sentences are treated as documents [3]- [5]. The state of the art and most successful models for sentence retrieval are the vector space model [1], [6], [7] and the language model approach with local context [1], [2]. The vector space model for sentence retrieval uses the term frequency -inverse sentence frequency method (TF-ISF) which is analogous to the traditional term frequency -inverse document frequency method (TF-IDF) used for document retrieval.…”

“…When it comes to the language model approach for sentence retrieval at first it was implemented as a trivial adaptation from document retrieval to sentence retrieval, with sentences treated as documents [7]. In the recent works language model approach was successfully modified to take into account the context of sentence, like the document which contains it or some number of previous or next sentences [1], [2], [8]. The TF-ISF method was also modified to take into account context, but unsuccessfully without statistically significant improvements [1].…”