Identifying and Extracting Quantitative Data in Annotated Text
D.J.M. Willems, H. Rijgersberg, Jan L Top · Socio-Environmental Systems Modeling · 2012
Abstract. In science it is difficult to reuse quantitative scientific data. For example, it is not possible to search for quantitative data in papers in a directed way, such as using the query "Select the storage modulus of dairy product A after the temperature has decreased from 90 to 4 ◦ C". This is caused by the fact that data is made available in (relatively) free formats as in scientific papers, spreadsheets, or databases, all with limited annotation and description of the way they were obtained. Meaning is lost, for example about what the numbers relate to (quantities and units are often poorly indicated). Many researchers, especially in the physical and computer sciences use LATEX in their creation of scientific papers. In this paper we present a set of LATEX-style files, which use the terminology defined in wurvoc.org, that can be used to annotate scientific papers. These style files define a set of commands, each representing a specific quantity or unit. If the LATEX is typeset into a PDF file, quantities and units in the PDF will be annotated with the appropriate references (URIs) to the corresponding concepts in the OM ontology. This will not only disambiguate the use of these quantities and units, but will also enable us to extract triples from the PDF, facilitating the use of SPARQL queries to answer advanced quantitative question. 1