A materialized view is a variation of the regular database view. Both are results of database queries. The regular (non-materialized) view is only logically stored in the database (i.e. the contents are not physically stored, but dynamically computed whenever needed). On the other hand, the results of the materialized view are physically stored as a concrete table in the database, and are retrieved directly when referenced.
Since materialized views are physically stored, they may become inconsistent with their original data sources (when the data sources change due to updates). For this reason, materialized views have to be periodically updated to account for changes in the base tables. This is not an issue with non-materialized views, since they are re-computed from the original data every time they are referenced.
Therefore, materialized views allow much more efficient access (since they do not have to be re-computed all the time), at the cost of some data being possibly out-of-date. They are most useful in data warehousing scenarios, where the base tables are not updated very often, and frequent queries of the actual base tables can be extremely expensive.
Materialized views can be used exactly in the same way as regular views. They can be used to simplify queries, and hide their details, or to store common computations that are referenced by multiple queries.
However, they have one additional use that is the quite the opposite of regular views. Recall that during query processing, queries that reference views are usually rewritten to reference the base tables. However, the query optimizer tries to reuse materialized views to evaluate queries. This is known as materialized view matching and utilization. Since the contents of these materialized views are already precomputed, using them for query evaluation can result in a significant performance improvement.