We found a match
Your institution may have rights to this item. Sign in to continue.
- Title
Gist: general integrated summarization of text and reviews.
- Authors
Lovinger, Justin; Valova, Iren; Clough, Chad
- Abstract
E-commerce is rapidly growing, with review Web sites hosting hundreds of reviews on average for any product. Reading so many reviews is tedious, time-consuming, and with the proposed Gist, unnecessary. We introduce Gist, a system to automatically summarize large amounts of text into informative and actionable key sentences. With unsupervised learning and sentiment analysis, Gist selects the sentences that best characterize a set of reviews. All of this is done in seconds, without prior adjustment or training. Gist extends the current state of the art with a modular system that can take advantage of a priori knowledge and adapt to new domains through easy modification and extension. Gist is a general framework, able to summarize any set of text and easily adapt to specific domains. A robust comparison with state-of-the-art summarization algorithms, on datasets containing hundreds of documents, proves Gist's ability to effectively summarize text and reviews.
- Subjects
ELECTRONIC commerce; WEBSITES; CONSUMERS' reviews; SUPERVISED learning; A priori
- Publication
Soft Computing - A Fusion of Foundations, Methodologies & Applications, 2019, Vol 23, Issue 5, p1589
- ISSN
1432-7643
- Publication type
Article
- DOI
10.1007/s00500-017-2882-2