This talk is an introduction to the first results obtained by the CHI-KNOW-PO project which studies medieval Chinese poetry using both computer-assisted text mining and close reading. Thanks to this methodology which pertains to digital humanities, it is possible to identify and make visible patterns in the use of words from different periods and authors. It becomes for instance obvious that feelings are expressed in a distinct manner during the Six Dynasties and during the Tang. It also allows us to see which words most frequently appear in cooccurrence and to suggest ways of reading the weaving of words pertaining to the semantic fields of the wind, autumn and sorrow for instance.