Basic Query Features

There are three elements to a query: where to search, what to search for, and what information to view.

Query Scoping

A query scope specifies the set of documents that must be searched. Typically scopes are specified by a directory path on a storage volume, such as d:\docs. IIS publishes Web "sites," which correspond to a virtual root that points to a collection of documents. This is illustrated in Figure 2.

Figure 2 - IIS Sites

"/Windows NT" and "/Applications" are two virtual roots on the server. The virtual roots map to physical storage locations on the local machine or another connected file server. Index Server indexes documents based on these virtual roots. An administrator can index all the virtual roots on a server or select a subset to index. Queries can be run against multiple virtual roots, against a single virtual root, or even against a single physical directory within a virtual root.

Query Restrictions

In addition to specifying the scope of a search by virtual directory, physical directory, and volume, Index Server allows query restrictions to be built by combining a number of "tests." Query restrictions available for use in any combination include:

• Searching for words and phrases.
• Searching for proximity of words or phrases to other words or phrases.
• Searching for words or phrases within textual properties. For example, searching for a word that appears in a document's abstract or summary.
• Searching for words or phrases within specific document types, such as within an Excel spreadsheet or a PowerPoint presentation.
• Using parameters of <, <=, =, =>, > against a constant, such as a date or file size.
• Using Boolean operators, AND, OR, and AND NOT.
• Using wild cards, such as "*", "?" and regular expressions for "fuzzy" queries.
• Ranking by quality of the match.

Free-Text Query

Index Server also allows free-text queries, in which the user can enter any set of words or phrases, or even a complete sentence as the query restriction. The system looks at the text and identifies all the nouns and noun phrases. If a complicated phrase were typed in, Index Server would extract two- and three-word combinations, as well as all remaining content words, while eliminating the "noise" words such as "a", "of", and "the".

For example, assume the user typed in the following free-text query:

"The Fulton County Grand Jury said Friday an investigation of Atlanta's recent primary election produced no evidence that any irregularities took place."

The system would identify the following words and noun phrases:

The words and phrases are combined into a restriction, weighted for proper ranking and posted as a query against the corpus. Free-text queries make querying very simple for the beginning user, because the user doesn't need to learn any type of query language.