Advanced search quick reference guide

When setting up a workspace in Relativity, admins need to consider what fields to search, which search indexes provide the most value, and how to optimize performance for the users, with minimal administrative overhead.

Note: This guide does not cover analytics indexes, nor will it detail the operators acceptable for use in each of these search engines.

Search type Keyword search dtSearch
How is it enabled? Relativity automatically indexes keyword searches when you load data into the system. The Active field should read Yes. (Search Indexes > Keyword Search) To access a dtSearch, you must first create a saved search. Search only on the Extracted Text field for optimal results. Next, used the saved search as the Searchable Set when creating a dtSearch index.
What can be indexed?

Available on all fields loaded into Relativity.

Except long text fields stored in Data Grid. In RelativityOne, extracted text is automatically stored in Data Grid.

 Available on all fields loaded into Relativity. See “Suggested Fields to be Indexed” below.
How is it used?

In the Documents tab:

  1. Click Add Condition.
  2. Click (Index Search).
  3. Select Keyword Search from the drop-down menu.
  4. Enter search terms.
  5. Click Apply.

See the Searching Quick Reference for more details on available search operators.

In the Documents tab:

  1. Click Add Condition.
  2. Click (Index Search).
  3. Select dtSearch from the drop-down menu.
  4. Enter search terms.
  5. If necessary, enable Fuzziness or Sorry for.
  6. Click Apply.

See the Searching Quick Reference for more details on available search operators.

Common search scenarios

Leveraging the above search index knowledge, use the matrix below to reference behavior across common search scenarios and learn suggested index tips.

  Keywords/Filters dtSearch
Engine SQL dtSearch
Noise words Yes Yes (customizable)
Search operators Search quick reference  
How to index Search quick reference  
When adding data (add new records) Automatically updates Incremental build
When changing existing data (overlay on existing records) Automatically updates Full build
When removing data (remove existing records) Automatically updates Full build
Suggested fields to be indexed Fixed length fields: Some long text fields with small amounts of text (ex: File Names) that are not indexed by dtSearch Index. Long text fields (ex: Extracted Text, Email To, Email CC.)
Suggested indexes N/A (not all fields flagged for indexing are grouped in an index.) • One for Extracted Text
• One for Email To, Email CC, Email BCC
Searching on individual fields Yes (select the individual field to search or filter on.) Yes (set up separate Indexes that index individual fields.)
Advantages • Instantaneous indexing
• Ability to search on individual fields		 • Ability to customize index
• Ability to search on individual fields; involves separate index setup
Disadvantages • Lacks specialized search capabilities
• Inability to customize indexes		 Manual index maintenance

**Only available on Data-Grid-Enabled Workspaces

Is Like and Contains operators on field level searching

  Is Like Contains
Behavior Wildcard (%) is applied to the front and back of the term. The field searches for the item entered.
Operators available None AND, OR, NOT, and Wildcard (%)
Multiple terms Terms entered on multiple lines are connected by an OR. Terms entered on multiple lines are connected by AND.
“Include in Text Index” Field does not need to be set to “Yes.” Only available for Fixed Length and Long Text Fields and needs to be set to “Yes.”
Comments Tends to run slowly. The best practice is to avoid running on large data sets. N/A

For example, you see the term “Valet Parking” appear the following ways using the various search operators listed below:

Term Term Term
“Valet parking” Exact phrase “Valet parking” Exact phrase “Valet parking”
Valet parking %valet parking% Valet AND parking
Valet park% %Valet park% “Valet” AND “park%”
Valet park* %Valet park% “Valet” AND “park*”
Valet park%% %Valet park% “Valet” AND “park%%”