Searching within a subset of a document base

Most of the sample queries provided with TEXTML Server are run against an entire docbase (see Searching for documents in a docbase).

This topic explains how to construct queries that are run against the results of one or more previously run queries. This allows you to improve performance by searching only within a subset of the documents in a docbase. It can also improve usability: your application can allow the user to search only within results that she has previously retrieved.

We'll look at three sample queries. The first query is SampleDateQuery.xml:

<?xml version="1.0" encoding="UTF-16"?>
<!-- SampleDateQuery.xml -->
<query VERSION="4.5" RESULTSPACE="Date Search Results">
    <key NAME="Date">
        <interval>
            <start INCLUSIVE="True">
                <date>
                    <year>2000</year>
                    <month>7</month>
                    <day>3</day>
                </date>
            </start>
            <end INCLUSIVE="True">
                <date>
                    <year>2000</year>
                    <month>10</month>
                    <day>1</day>
                </date>
            </end>
        </interval>
    </key>
</query>

It can search any docbase that has a Date index. It searches for documents within a specified date range. And it temporarily stores the results that it retrieves in a ResultSpace named "Date Search Results".

The second query is SampleMulticriteriaQuery.xml:

<?xml version="1.0" encoding="UTF-16"?>
<!-- SampleMultiCriteriaQuery.xml -->
<query VERSION="4.5" RESULTSPACE="Publication and Size Results">
    <andkey>
        <key NAME="Publication">
            <elem>Canada NewsWire</elem>
        </key>
        <property NAME="Size">
            <interval>
                <start INCLUSIVE="True">
                    <number>10000</number>
                </start>
                <end INCLUSIVE="True">
                    <number>19999</number>
                </end>
            </interval>
        </property>
    </andkey>
</query>

It searches for documents that meet two criteria: the publication must be Canada NewsWire; and the size of the document must be within a specified range. This query temporarily stores the results that it retrieves in a ResultSpace named "Publication and Size Results".

The third query is SampleCombinedResultsQuery.xml:

<?xml version="1.0" encoding="UTF-16"?>
<!-- SampleCombinedResultsQuery.xml -->
<query VERSION="4.5" RESULTSPACE="Combined Results">
    <andkey>
        <key NAME="FullText">
            <frq VALUE="2">
                <elem>business<anystr/></elem>
            </frq>
        </key>
        <include TYPE="ResultSpace">Date Search Results</include>
        <include TYPE="ResultSpace">Publication and Size Results</include>
    </andkey>
</query>

Because the third query uses <andkey>, it will only retrieve documents that meet all of the following criteria:

• The document must contain at least two occurrences of any word that begins with "business".
• The document must be in a ResultSpace named "Date Search Results".
• The document must be in a ResultSpace named "Publication and Size Results".

The effect is that:

• The third query combines the results of the two previous queries by ANDing together their ResultSpaces. This combined result is (normally) a subset of the docbase.
• The third query then searches in the FullText index of the combined result.