Classic rule-based machine translation systems heavily rely on rules for converting text from the source language to the target language. These rules, often created by linguists, generally work at the syntactic, semantic, and lexical levels. A classical rule-based machine translation system typically has three phases:
- The analysis phase
- The lexical transfer phase
- The generation phase
Illustrated in Figure 3.1 is a flow diagram of a typical rule-based machine translation system:
Figure 3.1: A flow diagram of a rule-based machine translation system
