Use this rule-based engine to extract data from tables.
The recommended engine to extract table data is the Brainware Table Extraction engine.
As with Brainware Table Extraction, traditional table analysis yielded several candidates, but the one with the highest confidence is used automatically. Therefore, no evaluation step is required. You can use scripts to access the remaining candidates.
In principal, there are two approaches to traditional table analysis.
- Fixed layout: If the position of table elements is constant, you can use their coordinates to locate the table and its elements. This resembles the zone analysis of text fields.
- Variable Layout: If the position of table elements varies, or to analyze different table layouts with a single set of settings, you can use column labels and formats to locate a set of table candidates. This resembles the format analysis of text fields.
You can also mix both approaches.
The Table Analysis Engine allows you to extract a single table from each document. This table can span multiple pages, provided that the table layout does not change over the entire range. To be accessible by table analysis, tables must consist of the following elements.
- Columns
- Rows
- Cells
Limitations
- Reading tables with interleaved columns is not supported.
- Reading tables with empty rows is only supported if the table has a footer line.