TextRunner: Open Information Extraction on the Web

Traditional information extraction systems have focused on satisfying precise, narrow, pre-specified requests from small, homogeneous corpora. In contrast, the TextRunner system demonstrates a new kind of information extraction, called Open Information Extraction (OIE), in which the system makes a single, data-driven pass over the entire corpus and extracts a large set of relational tuples, without requiring any human input. (Banko et al., 2007) TextRunner is a fullyimplemented, highly scalable example of OIE. TextRunner’s extractions are indexed, allowing a fast query mechanism. Our first public demonstration of the TextRunner system shows the results of performing OIE on a set of 117 million web pages. It demonstrates the power of TextRunner in terms of the raw number of facts it has extracted, as well as its precision using our novel assessment mechanism. And it shows the ability to automatically determine synonymous relations and objects using large sets of extractions. We have built a fast user interface for querying the results.