Summary
The Thesaurus function allows to define relationships between search terms, such as synonyms and antonyms. These settings affect how FactFinder assesses the similarity of records in the database.
Such similarities can only be assigned to individual words: "jeans" can be specified to be similar to "trousers", but cannot be defined as similar to "blue cotton trousers". To work with multi-word terms, Preprocessor should be used
Thesaurus is represented as an individual page tab for Search term rules page.
Adding Thesaurus entry
Thesaurus entries consist of word pairs (Source word and Target Word
) that have Relationship
to each other to an extent that is defined by a Direction
and similarity Devaluation
:
Default available Relationships, Directions and Devaluations are:
Synonym - bidirectional - 0% Devaluation
Soft Antonym - bidirectional - 100% Devaluation
Hard Antonym - bidirectional - Excluded from search result
Hard Antonym (uni) - unidirectional - Excluded from search result
Similar- unidirectional - 5% Devaluation
Very similar - unidirectional - 1% Devaluation
Generic term - unidirectional - 0.1% Devaluation
Expert - allows to manually set up Direction and Devaluation.
Tip: the direction from a pointless to a useful entry should be unidirectional. If both entries are useful, then choose bidirectional.
After entering Source word and Target word, selecting Relationship, Direction and Devaluation, adding optional Comment , new Thesaurus entry can be added by clicking Add Entry button
.
Managing Thesaurus entries
All created Thesaurus entries are listed in the grid together with their attributes.
The grid allows to filter entries:
hide/show entries deleted in this transaction
search for a specific entry
The following operations are available for transactions in the grid:
edit
clone to another channel
delete
bulk clone and delete operations are available for multiple selected (by
) entries
Any creation or change to an entry will mark it as locked
and require saving the transaction.
Transliterations setting can be activated to consider common transliterations for the matching, e.g entry “tür” → “tor” would also be triggered for the input “Tuer”.