Overview

A full-text index is a type of index specialized for efficient searching for string or text properties, especially in large text fields like descriptions, comments, or articles.

Full-text indexes work by breaking down the text into smaller segments called tokens. When a query is performed, the search engine matches specified keywords against these tokens instead of the original full text, allowing for faster retrieval of relevant results. Full-text indexes support both precise and fuzzy matches.

Showing Full-text Indexes

To retrieve information about full-text indexes in the current graphset:

// Shows all full-text indexes
show().fulltext()

// Show all node full-text indexes
show().node_fulltext()

// Show all edge full-text indexes
show().edge_fulltext()

The information about full-text indexes is organized into the _nodeFulltext or _edgeFulltext table. Each table provides essential details about each full-text index:

Creating a Full-text Index

You can create a node or edge full-text index using the create().node_fulltext() or create().edge_fulltext() statement. Note that each property can only have one full-text index. The full-text index creation runs as a job, you may run show().job(<id?>) afterward to verify the success of the creation.

System properties in Ultipa are inherently optimized for query performance and have built-in efficiencies. They do not support full-text indexing.

// Creates a node full-text index
create().node_fulltext(@<schema>.<property>, "<fulltextName>")

// Creates an edge full-text index
create().edge_fulltext(@<schema>.<property>, "<fulltextName>")

To create a full-text index named prodDesc for the property description of product nodes:

create().node_fulltext(@product.description, "prodDesc")

To create a full-text index named review for the property content of review edges:

create().edge_fulltext(@review.content, "review")

Dropping a Full-text Index

You can drop a node or edge full-text index using the drop().node_fulltext() or drop().edge_fulltext() statement. Dropping a full-text index does not affect the actual property values stored in shards.

A property with a full-text index cannot be dropped until the full-text index is deleted.

To drop the node full-text index prodDesc:

drop().node_fulltext("prodDesc")

To drop the edge full-text index review:

drop().edge_fulltext("review")

Using Full-text Indexes

To use a full-text index in filters, use the syntax {~<fulltextName> contains "<keyword1> <keyword2> ..."}:

• The ~ symbol marks the full-text index.
• The operator contains checks if the segmented tokens in the full-text index include all the specified keywords.
• Multiple keywords should be separated by spaces. If a double quotation mark appears in a keyword, prefix it with a backslash (\) to escape.

There are two search modes for full-text indexes:

• Precise search matches exact tokens to keywords.
• Fuzzy search occurs when a keyword ends with an asterisk (*), matching tokens that begin with the keyword.

Retrieving Nodes or Edges

To find nodes using the full-text index prodDesc where their tokens include "graph" and "database":

find().nodes({~prodDesc contains "graph database"}) as n
return n

To find nodes using the full-text index prodDesc where their tokens include "graph" or "database":

find().nodes({~prodDesc contains "graph" || ~prodDesc contains "database"}) as n
return n

To find edges using the full-text index review where their tokens include "graph" and those start with "ult":

find().edges({~review contains "graph ult*"}) as e
return e

Retrieving Paths

Using the ab() query to find paths within 5 steps, with the full-text index companyName applied to both the source and destination nodes:

ab().src({~companyName contains "Sequoia*"}).dest({~companyName contains "Hillhouse*"}).depth(:5) as p
return p

Note: Full-text indexes only apply to the first node in a Path Template or a K-Hop Template when retrieving paths.

For example, this query is not supported:

n().e().n({~prodDesc contains "graph"}) as p
return p

You may revise the query as follows:

find().nodes({~prodDesc contains "graph"}) as dest
n().e().n({_id == dest._id}) as p
return p

This query is not supported either:

n().e({~review contains "ult*"}).n() as p
return p

You may revise the query as follows:

find().edges({~review contains "ult*"}) as e
n().e({_uuid == e._uuid}).n() as p
return p