What does Table Confidence represent?

Table Confidence specifically refers to the minimum extraction confidence values for table cells. This metric indicates the reliability of the data extracted from each cell in a table format. It reflects how certain the extraction process is concerning the values identified and captured. The focus on minimum values underscores the expectation that even the least confident extraction should meet a certain threshold to be considered accurate.

In this context, recognizing that each cell in a table can have differing levels of confidence is crucial. While some cells might exhibit high confidence levels due to clear patterns or recognizable data, others may fall short. By using the lowest confidence level among all the cells, users can determine whether the overall data extraction from the table is trustworthy and usable. This approach enables a comprehensive understanding of the table's reliability rather than relying solely on averages or peaks, which might mask inconsistencies in data quality across rows or columns.

Other options, while they might relate to document data extraction in different ways, do not accurately define Table Confidence. For instance, considering average values would not effectively indicate the minimum confidence required for accuracy, while overall document quality might incorporate various factors beyond just table data. The highest confidence value within a table would not serve as a reliable gauge of how well the table as a whole performs since it could