LiDa is BnL’s first open linked-data project. This project, in collaboration with the National Literature Centre (Centre national de littérature, CNL), linked authors and documents from the collective catalogue of the bibnet.lu network to those from the Dictionary of Luxembourgish Litterary Authors (Autorenlexikon), a bilingual encyclopedia of authors having been active on the literary scene of Luxembourg since 1815.
How does LiDa work?
After a search on our platform a-z.lu, users have the possibility to explore the detailed bibliographic records. In each record, the author entry is linked directly to the CNL’s Autorenlexikon. This allows you to discover additional information on that specific author, such as biography, bibliographic and archival information, photos, literary prices and more. The linking works also in the opposite direction, for the end users starting their research in Autorenlexikon. On the main page of the author or on the individual document pages there are links to a-z.lu, giving the users the possibility to find and borrow the desired items from one of the libraries of the bibnet.lu network.
These links allow to easily navigate from a bibliographic record on the a-z.lu platform to the exact entry of a particular author on CNL’s Autorenlexikon platform. Similarly, the link in Autorenlexikon to the a-z.lu allows any user to easily locate a title and find a library in the network where they can borrow the item.
Get all LiDa links
All links displayed on the a-z.lu platform are available for download at the following URL:
Linking authors from two separate data sources (the collective catalogue of the bibnet.lu network and Autorenlexikon) is challenging due to the lack of common identifiers. For that reason, the linking algorithm is based on string matching techniques using the authors names and their bibliographical context, such as the title and publication year of the work.
Example of the rules used by the algorithm
– Use normalized forms for all strings (remove non-alphanumeric characters; replace characters with diacritics by their equivalents without diacritics; use uppercase characters).
– Eliminate definite and indefinite articles (in German, French and English) found at the beginning of the title (der, die, das, le, la, the etc.).
– Make no distinction between main author / contributor and secondary ones, pseudonyms or alternative names.
The validation of links between authors, respectively between titles was done using a multi-level accuracy system for links:
– Lower levels (author name(s) + title + publication year is identical in the two systems) can be automatically validated.
– Higher levels, which imply less matching accuracy (approximate match between author names or document titles), need to be manually validated.
The LiDa update process
The LiDa database contains both the links between authors and those between documents. The database is automatically updated on a regular basis, using exports from the collective catalogue and Autorenlexikon systems as input.
Different versions of the algorithm
1. The first version of the matching algorithm was implemented in 2013, after various improvements of the above-described steps.
2. At the end of 2014 – beginning of 2015, after much work
done by the BnL and CNL teams on the authority records of the the collective catalogue,
more than 15.500 matches were added to the LiDa database, taking advantage of
the existing links between bibliographic and authority records and previously
created links between authors in the collective catalogue and in Autorenlexikon.
3. Towards the end of 2018 – beginning of 2019, after the introduction of VIAF (The Virtual International Authority File) and ISNI identifiers into both the collective catalogue and Autorenlexikon, a complete rewrite of the algorithm will take place, using unique identifiers for the author matching.