Fig. 4: Schematic of table relation extraction algorithm. | npj Computational Materials

Fig. 4: Schematic of table relation extraction algorithm.

From: Automated pipeline for superalloy data by text mining

Fig. 4

Table direction (“by row” or “by column”) is first detected by estimating the row or column location of the target information in the table body, e.g., chemical elements. Alloy named entities in green cells and elements in orange cells are jointed by the row and column index of each table cell from corresponding sequences respectively, and finally written into a quaternary-tuple <article DOI, alloy named entity, property specifier(element), property value(content)>.

Back to article page