Pictographic matching
Mark A. Walch, Donald T. Gantz · 2004
In this paper, we introduce the concept of Pictographic Matching as a tool for document exploitation across multiple languages. The primary technology supporting Pictographic Matching uses graph-based pattern matching to detect the signature of words contained in images of documents. This signature takes the form of graphs created by the line forms used to construct written words. These graphs can be matched to known graphs representing alphabetic characters, groups of characters and character components. And, since graphs are intrinsic to the structure of written language, graph-based searching promises to support a common platform capable of handling a multiplicity of languages. The conceptual roots for Pictographic Matching are found in Graph Theory. Pictographic Matching is implemented through a multi-step workflow drawing upon a variety of analytical techniques. The objective of this paper is to provide a high-level overview of Pictographic Matching as it exists in its current implementation.