Query Tree and Query Expansion

When Exalead CloudView parses a query written in UQL, it is represented as a structured query tree, where the inner nodes are query operators and the leaves are Boolean predicates.

Query expansion generates a new, larger query tree by processing query expansion modules. This is known as the query rewriting step.

Query expansion modules process the query tree to:

  • enrich it, for example, by using synonyms and semantics,

  • interpret and normalize it, for example, by recognizing city names or acronyms.

In the end, the query expansion generates a new query tree. This is the query rewriting step.

Note: After this step, all Leaf and Rex nodes in the tree have been converted to FinalLeaf nodes.

This page discusses:

See Also
Configuring Query Expansion
Query Expansion Features
Enable query expansion

Query Expansion Example

The figure below shows the query tree for a user’s search for documents containing the words "Red Sox" or "Chicago bulls", and modified after January 1, 2016:

((Red Sox) OR (Chicago Bulls)) AND lastmodified>=2016/01/01
The Default Operator between Two Text Predicates (like "Red" and "Sox") is a Boolean AND.

Character Interpretation

Leaf nodes can have different prefixes such as:

Character

UQL interpretation

-

Excludes terms if - is at the beginning of a leaf.

The leaf is interpreted as a NOT context.

For example, Sarbanes-Oxley is considered as a single leaf, while Sarbanes -Oxley is considered as two leaves, one of which is negated.

+

The leaf is treated as exact, which disables some expansion operations.

., &, -,

and other word separators

interpreted as a NEXT operator instead of an AND, when used to separate words without additional white spaces.

For example, ASP.NET is parsed as ASP NEXT NET instead of ASP AND NET.

The query processor allows nonalphanumerical characters in words in a few special cases that you can configure.

Leaves can also have options. You must specify the options in brackets and separated by quotes.

The following leaf options are available:

  • query rewrite options, which modify the way the query processor works and expands the leaf.

  • raw options, which are passed to the index.

Matching Modes

Each word can exist in the index at several matching modes (or index kinds). The matching between word predicates and document words is defined by the matching mode of predicates. The matching mode can be one of the following:

  • Exact match: Matches only if the words match exactly. For example, The only matches with The. This is known as kind 0 (or k=0).

  • Case-insensitive or Lowercase match: Ignores case for matching. For example, the and The match. This is known as kind 1 (or k=1). This level can be specified by the "i" option in UQL.

  • Normalized match: Ignores case and accents, for example the, thé and Thé match. This is known as kind 2 (or k=2). This is the default matching mode but it can be changed using predicate options.

    Note: You can specify other matching modes in Linguistics > Advanced > Form indexing.