A preliminary step in microbial risk assessment in food is to gather and to capitalize experimental data in order to feed simulation models. We have designed, in the predictive modeling platform Sym'Previus (www.symprevius.org), a relational database, MicroRisk-RDB [1], to store experimental data in microbial risk. Data capitalization task encounters a challenging lock. Indeed, original data are spread out in heterogeneous data sources (scientific papers, technical reports / sheets, PhD thesis ...). Moreover, they are expressed in heterogeneous formats (mainly tables, text and graphics) and vocabularies. Manual entering of data in a database (e.g. MicroRisk-RDB) is therefore a time-consuming task. We present in this paper methods and tools to facilitate data capitalization and reusability in the framework of the Linked Data initiative , which promotes the publication and connection of structured data on the Web.