A natural language query consists only of normal terms in the user’s language, without any special syntax or format. ATG Search allows a user to enter terms in any form, including a statement, a question, or a simple list of keywords.
Keyword Queries
In a keyword query, the terms the user enters are used literally to retrieve any document that has all or any of those terms. The terms are not programmatically changed in any way, and match exactly only on themselves. Although effective, this mode is very restrictive. ATG Search supports this form of search through the Keyword query mode (see the Query Mode section in the Query Component Reference chapter), but the default behavior is to employ the strategies described in the Query Processing section later in this chapter. Users can also use query operators to achieve similar results (see the User-Entered Query Operators chapter).
Statement Queries
Statements, also called full sentence queries, are grammatical sentences (including questions) or sentence fragments. The statements may include low-content terms, such as articles (the), auxiliary verbs (would), or prepositions (of); these terms are also called noise words and are typically excluded from keyword queries, since they add little or no content to the query, and could exclude results that would otherwise relate to the content terms of the query. A search engine that handles this form of query typically uses one of three approaches:
Exclude low-content terms, which for this reason are also known as stop words. Since the low-content terms are ignored, they don’t impact the query results.
Require low-content terms, but weight them very low in terms of relevance. The terms are all required (a Boolean AND), but the effects of the low-content terms are diminished in terms of ranking the search results.
Treat all terms of the query as optional (a Boolean OR). Since the low-content terms are optional, they can’t exclude results, but will affect the results in terms of ranking.
ATG Search allows a combination of approaches. It provides a list of low-content terms, which can be either excluded from or assigned low weight in the query. This list can be edited through the Search Administration user interface. ATG Search does not require low-content (low-weighted) terms, but instead treats all terms of the query as optional. Thus, relevant results can not be excluded by a low-content term.
