In this paper, we deal with the problem of extracting and processing useful information from bibliographic references in Digital Humanities (DH) data. A machine learning tech- nique for sequential data analysis, Conditional Random Field is applied to a corpus extracted from OpenEdition site, a web platform for journals and book collections in the hu- manities and social sciences. We present our ongoing project with this purpose that includes the construction of a proper corpus and a efficient CRF model on this as a preliminary. This project is supported by Google Grant for Digital Hu- manities. A number of experiments are conducted to find one of the best settings for a CRF model on the corpus, and we verify them both in an automatic and manual way of evaluation.