Modifying or Using Advanced Index Parameters
To modify or use advanced index parameters
On the Coveo server, access the Administration Tool (see Opening the Administration Tool).
Select Index > Advanced.
In the Advanced page, modify or use the advanced index parameters presented in the following screen capture and table, and then click Apply Changes.
Determines the number of results subject to a second, more precise ranking—this second ranking insures that only the most pertinent documents are displayed on top of the result list. Values between 2 and 1,000 are accepted. The default is 100.
Note: This process requires considerable CPU resources. Increasing the value of this field above 100 documents can slow down the querying process. To speed up CES, it is possible to limit the extra ranking process to 20-50 documents. This is usually sufficient as the average user rarely displays more than two result pages.
The Commit transaction after n [Seconds (CES 7.0.6942+ (August 2014)), Minutes, or Hours] parameters specify at what time interval a transaction is committed even if it has not reached its maximum size limit (see What Is a Transaction?).
The default value is 1 minute. It is appropriate for normal index operations where existing sources are continuously refreshed, allowing new documents to be processed quickly, and therefore not limiting the index freshness.
When the searchability delay is an issue and near real-time indexing is enabled, you can consider setting a sub-minute value in seconds (see About Near Real-Time Indexing).
The main inconvenient of having short intervals is that crawlers are stopped shortly at each transaction commit. You should increase the value only while you initially build or rebuild very large sources.
Example: While you rebuild a 50 million document Exchange source, it is recommended to set a higher commit transaction time limit (ex.: 30 minutes) so that each transaction efficiently writes large amounts of data to the index.
Note: If a transaction is not written in the index, its content cannot be queried.
Determines if a majority (2/3) of the memory cache size is to be allocated for querying or indexing. It is also possible to allow equal resources (1/2 of the memory cache size) to each process. In which case, it is recommended to increase the memory cache available, so each process has access to sufficient memory.
Example: : When 50 MB of memory cache is available, and Optimize for indexing is selected, then 33 MB are available for indexing (2/3 of the memory cache) and 17 MB for querying (1/3 of the memory cache). When Optimize for indexing and querying is selected, 25 MB is devoted to each process (1/2 of the memory cache).
The value in the Memory cache size box determines the maximum memory size limit to force the commit of transactions.
Note: Increasing the memory cache available for indexing and querying slows down other system processes (ex.: index).
Provides equivalent names for fields (when CES is unable to find a field name in the CES7 database, it searches for alias sets). The Default Aliases set comprises alternate names for the built-in fields (see Available Field Aliases).
Example: author is an alias for sysauthor, because users tend to forget sys before field names (i.e. a query for @author returns the same results as a query for @sysauthor).
You can add new field aliases or create different field alias sets in the Field Alias Sets page (Configuration > Field Aliases) (see Managing Field Aliases).
Only one field alias set can be used on the index at a time.
CES 7.0.6547– (March 2014) You must restart the CES service to make a field alias set change effective.
Multiple Value Fields
A multiple value field is a field that can contain two or more values that are considered independent. When a multiple value field is used to create a facet in the search interface, its values are listed separately in the facet.
The Value Separators box allows you to change or add characters recognized as separators of independent field values. You can add more than one value separators.
The default character (;) should be used for all built-in fields. Other repositories such as databases or in-house applications could use different separators.
When you change or add Value Separators, a rebuild of all your sources is necessary for the modifications to take effect.
Changing or adding Value Separators characters can break existing field value separations. Contact Coveo Support for assistance before attempting to change this parameter.
Allows the usage of Wildcard characters in queries (see Search Prefixes and Operators).
In Number of Leading Chars, enter the minimum number of leading characters that the user must enter before a wildcard character. The default is 2. It is not recommended to use a value below two as this would result in computationally expensive wildcard queries.
In Number of candidates, specify the number of potential replacement terms for a wildcard expression. Increasing the value can affect performance. The default value is 32. You can safely increase this value to 128.
Note: Wildcard queries on the entire content of an index will always take longer to return results. For content where wildcards are appropriate, you should consider isolating the data in a field, and using the field for the wildcard search (see What Are Field Queries and Free Text Queries?).
Specifies the amount of memory to allocate to facet caching.
The calculation of the stem classes for queried words is automatically performed during off-peak hours to reduce the indexing time (see About Stemming). You can however initiate this calculation for all indexed terms at any time by clicking Perform advanced analysis on language stemming classes.
Example: After re-indexing large sources, you may want to not wait until the next off-peak period to update the stem classes.
Important: Forcing the calculation of all indexed terms by clicking Perform advanced analysis on language stemming classes may require significant computing resources.
Word Corrector Lexicon
The index maintains a word correction lexicon (WCL) that sorts indexed terms by their number of occurrences and is used by the spelling suggestion algorithm of the query correction feature to find more frequent spelling variants and propose a correction (see Query Correction Feature and How Are Misspelled Words Handled?).
The word correction lexicon calculation is performed during off-peak hours to reduce the indexing time. You can however initiate this processing at any time by clicking Rebuild Word Corrector Lexicon.