Reviews for Information Extraction

The average rating for Information Extraction based on 2 reviews is 3 stars.

Review # 1 was written on 2014-01-20 00:00:00 Shane Bailey This book is supposedly one of the few books available in the market that dedicate the entire book on Natural Language Generation (NLG). One usually only finds a section or a chapter (if at all) from books about Natural Language Processing (NLP) mentioning about NLG. As the title says, its main focus is on the system and its architecture. While there are different possible architectures for building such as systems, the book focuses on a three-stage pipeline architecture, with separate stages for what the authors call document planning, microplanning and surface realization. There were not too much discussion on the design part, e.g. how would one design the language output with different style or "personality". The book was supposedly written in mid-2000 but most of the citation seems to date back in mid 90's or earlier research. The approach used in the book is mainly rule-based and heavily rely on grammar. To the authors, a template-based is the simplest form of rule-based system. There were no discussions on using machine learning or statistical model. The example uses of language generation used in the book were mostly traditional language generation such as weather report, system documentation or personalized letter with no mention of a conversational system such as Siri. However, going through the book, one can learn about the challenges and methods to solve the language generation problems in the various stages. For LG, the most relevant section would be microplanning and surface realization as document planning is closer to the realm of the dialog manager. Throughout the book, there are many examples that range from picking up the right data, constructing the structure of the messages, determining the contents, leveraging grammar rules and syntactic constructs. The problem of lexicalization, aggregation and referring expression generation were discussed in detail. I think this book goes much deeper into the topics of language generation than we really need for LG and some of the concepts are hard to fully absorb with lots of citations of research and systems built from the academic community. However, it does give the readers the mindset when thinking about language generation and the problems that they will encounter. The book has lots of references and example systems that could be valuable for our LG Science effort if we want to dig deeper in any particular area.

Review # 2 was written on 2018-12-26 00:00:00 Anthony Dibattista یک کتاب اساسی و مفید. برای کسانی که به پردازش زبان طبیعی علاقه مند هستند بسیار توصیه می شود.