Automatic delineation and extraction of tabular data in portable document format using graph neural networks

A computer program product comprising:

• one or more computer readable storage media and program instructions stored on the one or more computer readable storage media, the program instructions comprising:
• program instructions to train a machine learning algorithm utilizing one or more supervised training methods with one or more training sets and one or more testing sets, the machine learning algorithm including a tree builder model and a cell builder model;
• program instructions to extract metadata corresponding to a set of tabular data in a table having border lines defining at least a body section of the table, the table being in a text-based portable document format (PDF), the metadata including relative positions of border lines making up the table and corresponding characters of the tabular data;
• program instructions to generate a first graph structure corresponding to the set of tabular data in the text-based PDF based at least in part on the metadata;
• program instructions to generate a first graph structure corresponding to the set of tabular data in the text-based PDF based at least in part on the metadata and the positions of characters and border lines in the table;
• program instructions to generate a vector representation comprising a set of feature vectors using the first graph structure;
• and
• program instructions to construct, by the trained machine learning algorithm, a tree structure corresponding to the tabular data based at least in part on the vector representation, the tree structure including a body node identified by the tree builder model based on the metadata associated with the border lines defining the body section of the tabular data.

A computer system comprising:

• one or more computer processors;
• one or more computer readable storage media;
• and
• program instructions stored on the computer readable storage media for execution by at least one of the one or more processors, the program instructions comprising:
• program instructions to train a machine learning algorithm utilizing one or more supervised training methods with one or more training sets and one or more testing sets, the machine learning algorithm including a tree builder model and a cell builder model;
• program instructions to extract metadata corresponding to a set of tabular data in a table having border lines defining at least a body section of the table, the table being in a text-based portable document format (PDF), the metadata including relative positions of border lines making up the table and corresponding characters of the tabular data;
• program instructions to generate a first graph structure corresponding to the set of tabular data in the text-based PDF based at least in part on the metadata;
• program instructions to generate a first graph structure corresponding to the set of tabular data in the text-based PDF based at least in part on the metadata and the positions of characters and border lines in the table;
• program instructions to generate a vector representation comprising a set of feature vectors using the first graph structure;
• and
• program instructions to construct, by the trained machine learning algorithm, a tree structure corresponding to the tabular data based at least in part on the vector representation, the tree structure including a body node identified by the tree builder model based on the metadata associated with the border lines defining the body section of the tabular data.