The Lexicon of Greek Personal Names (http://www.lgpn.ox.ac.uk/) started in 1972 as a British Academy Major Research Project to collect and publish all known ancient Greek personal names; it was well into digitisation by the early 1980s, so it has a proud 30 year history of involvement in technological developments and interest in long-term data management. LGPN primarily, however, produces printed volumes (seven to date) which cover different geographical areas, so the exposure of material in linked data form for SNAP necessitates doing some new work on mapping from the existing database structure to the new ontology.

The volumes published so far reference c. 250,000  individuals, with c. 36,000 distinct names from c. 4000 locations in the ancient world. The data are managed in a relational database with quite a complex structure (http://www.lgpn.ox.ac.uk/online/computerization/index.html) which already makes clear the distinction between a person, and a name as an entity in its own right (indeed, the Lexicon is essentially an onomastic project, not a prosopographical one). A mapping of this structure  to the Text Encoding Initiative Guidelines already provides a portable XML format for longer-term preservation and interchange,  and some work has been done on representing this as RDF/XML against the CIDOC CRM ontology, which has been exposed by the CLAROS project.

Our implementation for SNAP is managed as a transformation from the TEI XML format produced by the existing online search at http://clas-lgpn2.classics.ox.ac.uk/  (documented in Elaine Matthews and Sebastian Rahtz, The lexicon of Greek personal names and classical web services. (2013). Bulletin of the Institute of Classical Studies Supplement. 107-124), exposing the people and their names. Three issues arise in this work, relating to the underlying data:

1. The source for a name attestation is given as a textual bibliographical reference eg IG XII (7) 269, 2, which is not directly linked to the inscription online.  The references are also multiple, and secondary as well as primary (eg IG II2 672, 4; 2797, 4, 11; Ag. XV 79; = Tracy, Athens and Macedon p. 175, 5; Ag. XV 80 & Prak. Wilhelm p. 319; SEG XXXVIII 74, [5]; = PA 14395). This is not an easy problem to solve, and is unlikely to be possible within the scope of SNAP.
2. Relationships between people in LGPN are shown textually, and not resolved into links in most of the database. When we see “Φίλιππος (s. Ἀστυκρέων, f. Ἀστυκρέων), we do not record which Ἀστυκρέων are meant. This is largely resolvable automatically, as the person called Ἀστυκρέων is almost certainly from the same place, and the same approximate date, but this will take some careful checking.
3. The published volumes of LGPN do not show a unique ID for each person, but simply list people by region, by name and in chronological order. The reader seeing Αὐτοκλῆς cannot unambiguously locate him again in the online system or in SNAP.

As the Lexicon passes  its 40th birthday, projects like SNAP provide a very valuable opportunity to revisit data structures, consider how data will be managed in future, and see how its own work will be affected in coming years by the explosion of machine-readable data about the ancient world.