Representation of Lexical Networks and Word-Level Data Processing

The present chapter treats the kernel memory representation of various lexical networks and word-level data processing. For modeling language-oriented data processing, introducing the concept of serial-order detection, i.e., detecting a subsequent activation pattern among multiple symbolic units responsible for the respective concepts, is essential. As described in Chapter 3, such a serial-order detection can be represented by the pattern matching performed by a unit in the network of kernel memory. Also, the bidirectional activation transfer among the nonsymbolic and symbolic units enables various interactive processing relevant to the association between multiple concepts. Hence, it is said that the network structure comprising both the nonsymbolic and symbolic units is symbolically grounded. The description then proceeds to the kernel memory representation of word compoundings and morphologies of inflection and derivation. In addition, it deals with linguistic variables within the kernel memory context.