An Analyzer is a Lucene concept, representing a class that implements the Lucene org.apache.lucene.analysis.Analyzer abstract class.
Analyzers prepare text for indexing and searching. For example, it doesn't make sense that every word of a text field is indexed - stop words such as "and", "of" and "a" may be irrelevant during a search. If these are to be ignored during a field search then the field is tokenized, ie. passed through an analyzer before writing the field to the index and likewise for a term value being searched.
Analyzers are language-specific - what defines a word is not the same in all languages. Some can be configured to ignore common stop-words (an, the, if, etc), to ignore numbers, and so on. Analyzers used by the Generic Search Server are configurable on a per-Search Service basis.