Morphology refers to the derivation of a word’s root by analyzing suffixes and prefixes. This process is different from stemming, because it utilizes knowledge of the language and roots to drive a more accurate analysis.

ATG Search uses a practical definition of “root” as the base form of a word that is effective for retrieval. Since this is not necessarily the etymological root word, ATG Search calls this an index term to avoid any confusion. ATG Search can be inconsistent on the amount of morphology it does, if it results in improved search performance. For example, in general, ATG Search derives er forms down to their verb roots, such as attacker => attack+er, but some er forms are better left as index terms, such as user and computer.

There are two levels of morphological analysis:

A word can include both inflectional and multiple derivational endings.

ATG Search performs morphology on both the query and the documents. Doing morphology on both index and query terms greatly increases the matches between a query term and indexed documents, increasing retrieval results and improving the ranking of results. Alternatively, you could index by literal form and use morphology to expand the query terms into all possible forms, similar to the process described in the stemming section.

The default behavior of ATG Search is to reduce query terms down to their index terms and use those for retrieval. The Literal Constraint section of the User-Entered Query Operators chapter describes how to find only the literal form of a query term.

Note: Since ATG Search allows custom terminology, administrators can alter what ATG Search considers an index term, such as adding seller so that it doesn’t get analyzed to sell. See the Custom Terminology section of this chapter.

loading table of contents...