TOMATE: A heuristic-based approach to extract data from HTML tables

Extracting data from user-friendly HTML tables is difficult because of their different layouts, formats, and encoding problems. In this article, we present a new proposal that first applies several pre-processing heuristics to clean the tables, then performs functional analysis, finally some post-processing produce output. Our most important contribution regarding which address by projecting cells onto high-dimensional feature space in standard clustering technique used make meta-data apart cells. We experimented with two large repositories real-world our results confirm can extract them an F1 score 89.50% just 0.09 CPU seconds per table. confronted competitors statistical analysis confirmed its superiority terms effectiveness, while it keeps very competitive efficiency.