Searching with Lucene in DSpace:#
To integrate SKOS within DSpace, we added:
- indexation expansion: every concept reference is expanded in the Lucene search indexes:
- to all translations of the preferred term
- to all synonyms
- to all alias of the identification code (including the main concept identification code)
- to all notations in other coding scheme for the concept
- auto-complete to help the user select a term and provide the corresponding code to the search engine
- translation of the stored code for the concept reference, taking into account the user language
- faceted browsing
- improved result display and sort options.
We had to modify the tokenizers to ensure that a code following the syntax: ConceptScheme_ConceptId is indexed as one word and left untouched (no stemming for instance).
This work was presented at OAI7 in Geneva.
SolR, a future project?#
Making similar additions / adaptations /plugins to SolR would ensure that we can use SolR instead of Lucene (DSpace Discovery) and have a "pluggable" indexing engine with a known interface, powerful functions (facetted browsing, n-gram indexing, etc.) and nice user interfaces (BlackLight, ViewFind).
Any SolR based application would then benefit from ASKOSI benefits.