A comprehensive understanding of the molecular network underlying seed development regulations remains a major scientific challenge with important potential impact for fundamental research, agriculture and industry. Seed development requires the coordinated growth of different tissues that involves complex genetics and environmental regulations. Most of this knowledge is spread in thousands of articles. Integrating knowledge conveyed by text with knowledge derived from experimental data requires a common model in view of modeling in systems biology. In this work, we focused on a model organism, Arabidopsis thaliana. We designed a knowledge model that meet the needs of text-mining, (i.e. manual annotation of texts and automatic information extraction), experimental data indexing and retrieval and reuse to other plant systems. The model defines 6 sets of entity types that cover 16 types of biological entities and external factors. The first two sets are central in regulation description: genetic process entities and their products, such as molecular entities or biochemical processes. The last four, Genotypes, Tissues, Development phase and Environmental factor, define the observed conditions for the regulation. Events represent binary and n-ary relations between entities: three event types describe regulations involving gene expression, metabolic pathways, regulatory networks and molecular accumulation, which is critical in seed development. Additional qualifiers specify involvement, activation, inhibition or requirement. Molecular interactions are represented by three biological events: Binds_to, Encodes and Interacts_with. Comparison, Belongs_To, Found_In and Found_During events are used to describe similarity, membership, spatial and temporal relations among entities. All events have Negation and Speculation modalities. This model was used by biology experts to manually annotate a reference corpus and train our Alvis methods that have shown its relevance.