Visual information extraction

Aggregated search

“…In particular, HTML tables and HTML lists are known to contain relational data. -Decoration, visual appearance [Aumann et al 2006;Meng et al 2003;Yoshinaga and Torisawa 2007]: Sometimes the structure of a document is easier to learn through its visual aspects, especially when a pattern in terms of tags is difficult to define or learn. -PMI and search hits [Church and Hanks 1989;Turney 2001;: Pointwise Mutual Information (PMI) is a statistical measure that indicates possible correlation between two expressions.…”

Towards domain-independent information extraction from web tables

“…Zhao et al [41], Zhai and Liu [40] and Simon and Lausen [28] describe different approaches for detecting repetitive patterns on web pages, which are predominantly source-code based and enhanced with visual cues. In contrast, Aumann et al [3] describe a system that works only on a hierarchical structure of the visual representation (experiments are performed with PDF documents) and learns to recognize text fields such as author or title from manually tagged training sets of documents. Conversely, our approach does not attempt to find individual text fields, but rather, larger structures, does not require training sets and neither imposes a tree structure on web pages.…”

“…natural form by discovering and interpreting patterns in data [2]. It can be applied to textual data [37,42], numeric data [40,41], spatial data [43,45], web data [39] and visual data [38] as well.…”

The Soar of cognitive architectures

“…natural form by discovering and interpreting patterns in data [2]. It can be applied to textual data [37,42], numeric data [40,41], spatial data [43,45], web data [39] and visual data [38] as well.…”