Document Structure Analysis

The image shows the result of an automatic structure analysis on papers. The legend at the top shows the mapping of the different logical structures to colors. The center view is used for editing and correcting the structure annotation of a single page. The thumbnails of several pages are shown on the left, providing a quick overview of the analysis result. This way also errors of the automatic methods can be spotted. For instance, the first thumbnail in the third row has a table in the upper left but a caption of the table is missing.

◀ Back to Text Visualization Overview

The logical structure of a document describes the document’s logical components. Examples for logical components are headlines, tables, footnotes, or enumerations. Several applications in the document processing domain benefit from the information of the logical structure, for instance information extraction or document retrieval tasks. Unfortunately, the typical document formats do not contain all the structure information desirable for automatic document processing. In order to make use of the document structure, the document has to be analyzed and the original structure of the document has to be reconstructed.

The goal of the project is to develop a general framework for automatic structure analysis that is applicable to different document types, e.g. file cards, business letters, or papers. The combination of visual and automatic methods allows an efficient and effective creation of the required document models according to the user’s need.

More information about this and related work can be found in the following publications.