In this chapter, we discuss how to prepare the foundation for the running of the TransRecipe automatic translation system. The lexical database of a computer translation system, as we know, is the basis on which other parts of the system, such as rules and examples, are built. Without a substantial database, no system can work effectively. As what comes out from the computer is based on what has been put in, the quality of translation equivalents in the database is a major issue that should not be neglected. If the translation of terms is poor, obviously the output will be poor. The role of a translation expert in creating an effective translation system cannot be overemphasized. Engineers tend to focus, among other things, on the speed of text generation to enhance translation productivity. Linguists are more concerned about structural differences of the source and target languages, while translators want to ensure that the output is comprehensible and acceptable.