Abstract:Semantic information comes from the things being recognized and understood gradually, and thus it is often in evolution during the modeling process. Existing semantic models usually describe the objects and the relationships in an application-oriented way, which is unsuitable to reuse the schemas during the semantics evolution. In this paper, we propose a new graph-based semantic data model to overcome the limitation. SemGraph adopts a meaning-oriented approach to specifying the subjective view of the things and uses the certain meta-meaning relationships to build a graph-based semantic model. The model is simple but expressive, and is especially fit for the semantics evolution. We introduce the basic concepts and the essential mechanisms of the model, demonstrate its features with examples and typical cases of semantics evolution.

2 Motivation 2 . 1 A scenario of semantic data modeling