Hide Comments
Hide Comments

Comments (0)

The attribute full-text search is used for the configuration of a searchindex for the full-text search.

 

icon_comment

The full-text search web application has to be installed in order to continue with this chapter. Make sure you have read the chapter WebOffice FTS-Index and installed the required components.

Theoretically it would be possible to use one FTS-Index in more WebOffice 10.9 SP1 applications but VertiGIS DOES NOT RECOMMEND this scenario! ONE FTS-Index application should be used by ONE WebOffice 10.9 SP1 application. See also the chapter Set up several Cores in WebOffice FTS-Index for a full description of how to setup more FTS-Index applications.

 

Full-text search configuration

Full-text search configuration

 

Property

Description

Searchindex address

URL of the search index application for the full-text search, e.g. http://localhost:8983/FTS-Index/#/WebOffice. The name of that application is also the name of the application context descriptor file (e. g. FTS-Index.xml) without file extension.

icon_cross-reference

For more details on installation see chapter Installation of WebOffice FTS-Index.

Username

Username for accessing WebOffice FTS-Index using Basic Authentication.

icon_cross-reference

Default user name is 'weboffice'. The user name can be changed - see therefore chapter Configuring WebOffice FTS-Index.

Password

Password

icon_cross-reference

Default password is 'weboffice4ever'. The password can be changed - see therefore chapter Configuring WebOffice FTS-Index.

Index re-creation time

Defines when the search index (i.e. all defined subindices) shall be automatically recreated. The time value format is HH:MM, multiple time values are separated with a semicolon (';').

If no time value is defined, there will be no automatic creation of subindices. Alternatively, subindices can be created using the SynAdmin-page or Batch Administration Request.

Index re-creation type

Defines which subindices will be re-created.

ALL_SUBINDICES --> All defined subindices will be re-created.

ALL_OUTDATED_SUBINDICES --> Only those subindices that have the status 'out of date' at the given point in time will be re-created.

icon_comment

The index re-creation type is only taken into account if the attribute 'Index re-creation time (indexing_time)' is set. If no time value is defined there, no scheduled creation of subindices will take place.

Delay commit?

The Creation of a subindex is finalized through a COMMIT-mechanism. The search index will only be updated with the newly indexed data when the COMMIT is performed.

Due to this mechanism, the (old) indexed data will be available for searching while a subindex is calculated anew.

This may lead to subindices temporarily being available in a half-finished state only, if several subindices are created simultaneously (multi-threading, see parameter "Init-Threads number").

With the value 'true' (default is 'false') subindices are only made available after they are completely created.

While the subindex is calculated, the old version (insofar present) can still be searched.

However, the update to the new version might happen with some delay, as this can only be done after the last of the simultaneously created subindices has been finished.

Max. timeout (s):

If a request to FTS-index is not answered within the duration configured here, the connection is closed with an error.

Particularly when many or very large subindices are used, FTS-index may be busy for long durations (e.g. for COMMIT or OPTIMIZE request) and subsequent responses are delayed. It may then be necessary to see a higher timeout threshold.

Default value: 60, i.e. 1 minute

icon_cross-reference

For large amounts of data, it may be necessary to increase the Timeout for connections to the FTS index, see HTTP Client Settings.

Init-Threads number

Maximum number of concurrent threads for the initialization of subindices, i.e. the number of feature classes or tables which can be indexed in parallel. A high value may put too much load on the server and increase the total time needed for the initialization. Each init-Thread causes a REST request at ArcGIS Server.

Init-block size

Maximum number of features that are requested from the ArcGIS Server for the initialization in one single block.

If a layer contains more features, multiple requests are sent to the server. if the server returns less features (maxRecordCount) than defined here, that amount defines block size.

A large block size leads to a temporarily high memory consumption during the initialization and may lead to an OutOfMemory.

Maximum indexing duration

Duration in seconds after which the indexing of a subindex (i.e. a feature class) is aborted. Default value: 7200, i.e. 2 hours

Maximum latency

Maximum time that is waited in total when a connection problem occurs while access a service (i.e. feature class). A new connection attempt is started every 3 seconds. Default value: 120, i.e. 2 minutes

Fuzzy search factor

A value between 1.0 and 0.0 which describes the fuzziness of matching a search term with a result when the search assistant fuzzy? is used. The value "1.0" indicates a complete match (very strict, not recommended). The value "0.01" matches every result with almost any given search terms (very fuzzy, not recommended).

Phonetic alignment active?

Specifies whether a phonetic alignment shall be applied for the full-text search.

Refer to the following attribute 'Phonetic alignment' which defines the alignment that may be applied.

Phonetic alignment

Defines a comma-separated list of character substitutions that are used when indexing subindices or sending a full-text search request. These substitutions can be used to configure a phonetic search. Some examples (applicable to the German language):

umlaut "ä" = "ä->a,ae->a"; long vowel "i" = "ieh->i,ie->i" etc.

The reduced text leads to a higher probability to find a result with a phonetic instead of the correct spelling. Only lower case characters are used for the Phonetic alignment.

icon_warning

CAUTION!

The order of the given substitution blocks is important.

Searchindex language

The locale (two-letter language-code, according to ISO 639-1) that is used to build the search index. However, data in other languages can be indexed as well. The locale is used to transform special characters (e.g. ß, Ž, Ç) for the phonetic alignment, as the full-text search is case-insensitive.

Possible Values: ar, de, en, es, fr, sk, sl, ru and many more (over 100 languages).

Word delimiters

The full-text search is based on word blocks.

The ' ' (space character) is always interpreted as word delimiter, even if it is not configured here.

Depending on the given data, additional delimiters (e.g. '-', '#', '+' or '/' etc.) can improve the search result.

The characters may be configured here in one string (e.g. like "-#+/", i.e. without any separation).

The default setting is " -", that is to say the space character and the minus character.

The character '*' (asterisk) cannot be used as word delimiter, because of its special function as Wildcard character.

The character '~' (tilde) cannot be used as word delimiter, because of its special function in the search mode "Fuzzy Search".

Search assistance exact?

The value "Yes" (default) activates the "Search assistance exact".

The activation of the "Search assistance exact" provides exact search using the given search input criteria.

icon_comment

Until a search result is reached, the activated search assistants are being used in the following order:

- Search assistance exact

- Search assistance wildcard

- Search assistance fuzzy

- Search assistance LIKE

As soon as one of the search assistants finds any results, these are presented in the client. Any potential subsequent search assistants are skipped.

If all search assistants are deactivated, the search assistant 'wildcard' is being used per default.

Search assistance wildcard?

The value "Yes" (default) activates the "Search assistance wildcard".

The activation of the search assistance increases the chance of a search hit. When the given search term does not return any results/suggestions doing an exact search or fuzzy search, the search term is automatically adjusted and the search runs again a 3rd time (automatically). The "Search assistance wildcard" adds a trailing "*" (=wildcard) to the given search term. The value "true" (default) activates the "Search assistance wildcard". It will be used after the "Search assistance fuzzy" (provided that it is activated).

icon_comment

At least one of the four assistances ("exact", "wildcard", "fuzzy" and "LIKE") has to be enabled ("Yes"). If all four are disabled (set to value "No"), WebOffice 10.9 SP1 will assume "Search assistance wildcard"="Yes" by default.

Until a search result is reached, the activated search assistants are being used in the following order:

- Search assistance exact

- Search assistance wildcard

- Search assistance fuzzy

- Search assistance LIKE

As soon as one of the search assistants finds any results, these are presented in the client. Any potential subsequent search assistants are skipped.

Search assistance fuzzy?

The value "Yes" (default value is "No") activates the "Search assistance fuzzy".

The activation of the "search assistance fuzzy" increases the chance of a search hit. When the given search term does not return any results/suggestions doing an exact search, the search term is automatically adjusted and the search runs again a 2nd time (automatically) in "Fuzzy" mode. The "Search assistance fuzzy" uses the fuzzy search mode. The fuzziness can be defined in the attribute "Fuzzy search factor".

icon_comment

Until a search result is reached, the activated search assistants are being used in the following order:

- Search assistance exact

- Search assistance wildcard

- Search assistance fuzzy

- Search assistance LIKE

As soon as one of the search assistants finds any results, these are presented in the client. Any potential subsequent search assistants are skipped.

If all search assistants are deactivated, the search assistant 'wildcard' is being used per default.

Search assistance LIKE?

The value "Yes" (default value is "No") activates the "Search assistance LIKE".

The activation of the search assistance increases the chance of a search hit. When the given search term does not return any results/suggestions doing an exact search, fuzzy search or wildcard search, the search term is automatically adjusted and the search runs again another time (automatically).

The "search assistance LIKE" adds a leading and a trailing '*' (=wildcard) to the given search term, e.g. search for "*searchtext*".

icon_warning

CAUTION!

A search with wildcards needs more resources than a conventional search. Additionally, it carries the threat of producing too many (and therefore useless) results.

icon_comment

Until a search result is reached, the activated search assistants are being used in the following order:

- Search assistance exact

- Search assistance wildcard

- Search assistance fuzzy

- Search assistance LIKE

As soon as one of the search assistants finds any results, these are presented in the client. Any potential subsequent search assistants are skipped.

If all search assistants are deactivated, the search assistant 'wildcard' is being used per default.

Combined Suggestions?

The value "Yes" (default is "No") enables a special wildcard search for suggestions.

For each search value a separate search is executed, one time with and one time without a wildcard.

The results of all these searches are combined and sorted by highest returned search score.

This option only applies for suggestions, not for search results. If activated, any other search assistants for retrieving suggestions are deactivated.

This option is recommended for searching with multiple search values in a data set with a high number of identical values (e.g. address sets).

Separator for suggestions?

Use parameter "Object name concatenator" in Result Field List for separating suggestions and multilayer object result of full-text search. If "false" a whitespace (" ") will be used to separate the field values.

icon_comment

If this parameter is set to "Yes" the object name concatenator of the first result field list of the first search result on the respective layer of the first relevant project will be relevant.

Every time the configured concatenator is changed, the existing subindex is "out of date" and has to be recreated.

Date format

Defines how field values of type DATE are formatted for the full-text search index.

On the one hand this format determines how date values should be entered for the full-text search and on the other hand this format determines in which way date values are displayed within the object names. The order of the numerical values for day (DD), month (MM) and year (YYYY) as well as other delimiters can be configured. A possible time is always formatted as HH:MM:SS.

Support all field types?

In older versions of WebOffice it was only possible to use some few field types for the full-text search. If this attribute is set to "false", the old behavior can be restored (backwards compatibility).

icon_comment

Currently, these field types can be used:
Regarding search fields:
GLOBALID, GUID, OID, SINGLE, DOUBLE, DATE, INTEGER, SMALL_INTEGER, STRING

Regarding result fields:
GLOBALID, GUID, OID, SINGLE, DOUBLE, DATE, INTEGER, SMALL_INTEGER, STRING

In older WebOffice versions these field types could be used:
Regarding search fields:
INTEGER, SMALL_INTEGER, STRING

Regarding result fields:
GLOBALID, GUID, OID, DOUBLE, INTEGER, SMALL_INTEGER, STRING

Configuration of the full-text search

 

icon_cross-reference

Also see chapters Synonyms for Full-Text Search and Full-Text Search Language Settings for further full-text search configuration options in the application configuration.

More Information about the usage of Full-Text Search in the SynAdmin, possible states and supported field types of an index is shown in the chapter Full-Text Search.

When encountering problems related to the full-text search, follow the troubleshooting described in chapter Problems Related to Full-Text Search.

 

Comments (0)

Ⓒ Copyright 2022 by VertiGIS GmbH