Application developers who work on catalogs, digital libraries, and knowledge repositories are no doubt familiar with information retrieval, or IR. An IR application takes in user-supplied criteria and searches for the items or documents that best match the intent of the user. This is one of the major ways that IR differs from standard SQL queries, which either do or do not find matches for the query criteria. Good IR systems can help determine what documents are about, and return those documents that are most relevant to the search, even if they don’t match the search exactly.
Perhaps the most challenging task in IR is to support indexing and querying in multiple languages. English, for example, is a single-byte language that uses whitespace to separate words. Retrieval of information is substantially different when working with Japanese, which is a multibyte character set that does not use whitespace as delimiters.
Oracle Text, an option available in both the Oracle Enterprise Edition and the Oracle Standard Edition, provides full-text IR capabilities. Because Oracle Text uses SQL for index creation, search, and maintenance operations, it works very well in PL/SQL-based applications.
Called ConText and interMedia in prior releases, Oracle Text really came of age as an information retrieval solution with Oracle9i Database. With Oracle Text:
All NLS character sets are supported.
Searching across documents in Western languages, as well as in Korean, ...