ABSTRACT

This chapter describes the SCALIR "retrieval engine." In particular, it explains how the system's internal task of retrieving relevant documents is realized as a process of spreading activation. This explanation ultimately rests on the short-term dynamics of the system, where short-term means during a single query, as opposed to the long-term changes (learning) that occur over many retrieval sessions. The view presented here is necessarily narrowly focused. In particular, almost all discussion of how users actually engage in the retrieval process is deferred until Chapter 10.