We present in this paper a model of personal knowledge representation for lifetime storage. In the model we separate the knowledge layer from the resource layer. The knowledge layer consists of a network of atomic knowledge units situated in space and time. Re- sources are data packages (bit sequences) that can be ren- dered by some device into any human-perceivable form. The two parts complement each other: the knowledge network can be seen as annotations of the resource base (multimedia store) while resources can serve as means for the interpreta- tion of knowledge units as well as a way to index and access them. For the knowledge network we propose a simple formalism that we consider could support the emergence of a language capable of describing increasingly complex situations of the real world and, by time, to represent any information that is expressible by natural language.