Large amounts of textual data related to the agriculture domain are now available. Knowledge discovery becomes a crucial issue for research organizations, decision makers, and users. Our work investigates the use of \emph{Text Mining} methodologies in order to tackle several issues such as Animal Disease Surveillance, Open Data in Agriculture Domain, Information Extraction from Experimental Data. In this context, we have defined a new Knowledge Discovery from Texts (KDT) process applied to the agriculture domain (http://textmining.biz/agroNLP.html). This one is divided into four steps: (i) data acquisition, (ii) information retrieval, (iii) information extraction and disambiguation, (iv) visualization and evaluation. In this KDT process applied to specific use-cases, the integration of expert knowledge has a key role.