Integrating spreadsheet data via accurate and low-effort extraction

Abstract

Spreadsheets contain valuable data on many topics. However, spreadsheets are difficult to integrate with other data sources. Converting spreadsheet data to the relational model would allow data analysts to use relational integration tools. We propose a two-phase semiautomatic system that extracts accurate relational metadata while minimizing user effort. Based on an undirected graphical model, our system enables downstream spreadsheet integration applications. First, the automatic extractor uses hints from spreadsheets' graphical style and recovered metadata to extract the spreadsheet data as accurately as possible. Second, the interactive repair identifies similar regions in distinct spreadsheets scattered across large spreadsheet corpora, allowing a user's single manual repair to be amortized over many possible extraction errors. Our experiments show that a human can obtain the accurate extraction with just 31% of the manual operations required by a standard classification based technique on two real-world datasets.