Sampling strategies for information extraction over the deep web

详细信息    查看全文

• 作者：Pablo Barrio ; ^{pjbarrio@cs.columbia.edu} ; Luis Gravano ^{gravano@cs.columbia.edu}
• 关键词：Information extraction ; Sampling ; Deep web ; Text mining ; Scalability
• 刊名：Information Processing & Management
• 出版年：2017
• 出版时间：March 2017
• 年：2017
• 卷：53
• 期：2
• 页码：309-331
• 全文大小：2306 K
• 卷排序：53

文摘

First large-scale and fine-grained evaluation of query-based sampling techniques. Learned keyword queries perform substantially better than queries derived from tuples. Focusing on—and processing exhaustively—effective queries leads to high efficiency. Focusing on—and processing in rounds—less-effective queries favors quality. Filtering underperforming queries favors sampling efficiency but hurts quality.