Advanced search via lucene syntax

[Twan Goosen]

The VLO search box should support (limited?) queries using the lucene syntax so that users can perform more advanced searches over the full text index.

See ​https://wiki.apache.org/solr/DisMaxQParserPlugin#Query_Syntax for a description of the support in Solr.

[Twan Goosen]

As of VLO 3.3-SNAPSHOT r6284 the ​DisMax query parser is used. This already gives support for things like

• exclusion: corpus -hugo (vs corpus)
• phrases: "north america"' (vs north america)

We should probably switch to (or allow switching to) ​ExtendedDisMax which has support for AND, OR, ...

[Twan Goosen]

Extended dismax used as of r6300
TODO: add some form of usage instructions/documentation

[Twan Goosen]

Some examples that currently work on the ​testing version:

• German AND acquisition (notice collections on top)
• Berlin AND (german OR deutsch*)
• modality:*speech* OR modality:*spoken* (notice modality facet)
• (modality:*speech* OR modality:*spoken*) AND German AND Dutch
• genre:discourse AND Turkish AND (Dutch OR German)
• languageCode:"code:swe" AND country:Finland
  • After potential addition of ‘language’ search field: language:Swedish AND country:Finland

[Twan Goosen]

Consider adding field aliases for some fields with unfriendly names.

Candidates:

• languageCode -> lang
• languageName -> language (#770)
• resourceClass -> resourceType, type (though currently not queryable)

edismax documentation of the feature that should allow for this: ​http://wiki.apache.org/solr/ExtendedDisMax#Field_aliasing_.2BAC8_renaming

[Twan Goosen]

Alias 'language' for 'languageName' was added, so you can now also do, e.g.,

modality:*spoken* AND language:German AND language:Polish AND NOT language:English

[Twan Goosen]

NOTE: For the instructions: ​tooltips functionality of JQuery UI can be used, see this ​post for instructions on how to include links (for sample queries)

[Twan Goosen]

Added syntax documentation and a link to it with a JS tooltip in [6337:6345/vlo/trunk/vlo-web-app]

[Twan Goosen]

In current trunk version, the language field search examples with wildcards do not work, and it's also case sensitive. Probably best to make (a copy of) language name a text field instead of String.

[Twan Goosen]

Replying to twan.goosen@…:

In current trunk version, the language field search examples with wildcards do not work, and it's also case sensitive. Probably best to make (a copy of) language name a text field instead of String.

works with current schema.xml (currently importing on ​http://catalog-clarin.esc.rzg.mpg.de/vlo).
Updated one erroneous example in r6406