Table Header Detection via Global Orthogonal Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing header detection technologies are inadequate for complex tables with embedded headers and those converted from other files, as they typically rely on analyzing features of single rows or neighboring rows without considering the entire table context, and assume headers are in the first row, which is not always the case.
Innovation Solution
A method and system that determine headers in a 2D array or table by calculating scores for each row or column based on pairwise comparisons of cells across orthogonal columns or rows, comparing these scores to identify which rows or columns include headers, using a global machine learning approach that considers the entire table structure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional header detection technology analyzes only single row or neighboring row features, then the detection process is simple and fast, but the accuracy is insufficient for complex tables with embedded headers and converted files
Solution Approach 1:
The patent transitions from analyzing single rows or neighboring rows (one-dimensional approach) to analyzing the entire table structure (two-dimensional approach). By computing scores for all rows and columns and comparing them globally, the system captures table-wide patterns and relationships that local analysis misses, thereby improving header detection accuracy for complex tables while accepting increased computational complexity
Solution Approach 2:
The patent creates a universal header detection method that works across multiple table types (simple tables, complex tables with embedded headers, and converted files) by using a unified scoring approach based on global table analysis. The same score computation and comparison mechanism handles diverse table structures, making the detection system adaptable to various scenarios without requiring separate specialized algorithms
2Ease of operation
If header detection assumes the first row contains headers, then the detection is straightforward, but it fails to identify headers in tables where headers are not in the first row
Solution Approach 1:
Instead of assuming the first row contains headers and verifying this assumption, the patent inverts the approach by computing scores for all rows and determining which row has the highest score. This inversion allows the system to identify headers based on their actual characteristics within the table structure rather than their positional assumption, significantly improving reliability for tables with non-standard header placement
Solution Approach 2:
The detection system allows the table data itself to determine the header location through score computation and comparison, rather than relying on external assumptions or predefined rules. The row with the highest score automatically identifies itself as the header row based on its intrinsic characteristics relative to other rows, making the system self-determining and adaptable to various table formats
Data Source
AI summary
A method, system and computer-usable medium for detecting headers in various documents, such as PDF and HTML files. The files are converted to a two dimensional array or table, having orthogonal rows and columns. Either rows or columns are determined to include headers. For determining if rows include headers. For each row in the array or table, pair wise comparison is performed for each cell of each column that is orthogonal to that row. The pair wise comparison scores or values are summed up for each orthogonal column to that row and the sum across for all the orthogonal columns to row provide a score or value for that row. Row scores are evaluated relative to one another to determine likelihood of headers in the row. For determining if columns have headers, similar calculation is performed between columns and their orthogonal rows.