Quality Control

As described above, extracting the personal names from Trismegistos Texts is a work in progress, and because of the varying methods, we are at different stages for each language and even for each text. We distinguish the following phases:
phase 0: not yet done, personal names still to be added
phase 1: some personal names added, but possibly incomplete
phase 2: all personal names supposedly added, to be checked
phase 3: all personal names added, first check done
phase 4: intertextual prosopographical identifications investigated

Phase 4 has not yet been reached, but even phase 3 is rare, as the following table illustrates:

  Phase 0 Phase 1 Phase 2 Phase 3
Greek papyrology 23% 1% 76%
Demotic (Pap. & Epigr.) 14% 15% 36% 35%
Greek Epigraphy 18% 6% 76%
Coptic Papyrology 2% 98%
Latin Papyrology 100%
Latin epigraphy 35% 0.5% 64.5%
Coptic epigraphy 100%
Hierog. & hierat. papyrology 34% 62% 3% 1%
Hierog. & hierat. epigraphy 41% 58% 0.5% 0.5%
Aramaic (pap. & epigr.) 100%
Arabic papyrology 100%
Please note that the 23% in phase 0 for Greek papyrology are mainly post fifth century texts and new texts not yet entered in Duke in October 2008. Excluding these categories, the coverage in phase 2 is about 95%.

Amongst the inaccuracies and mistakes to be corrected in phase 3 are incorrectly recognized cases or incorrect accentuation for Greek, or unstandardized transliterations for Egyptian. The addition and correction of column and line numbers is a desideratum for all references, while the one-to-many and many-to-one relations between documents in Trismegistos and the DDbDP have also caused the listing of names belonging to the wrong texts. Please also note that in processing the DDbDP material, for technical reasons we were unable to add indications of abbreviations ‘( )’ , damages ‘[ ]’ or other diacritic signs. This also should be corrected in due course, but the direct link to the full text in the Papyrological Navigator or in the Packard Humanities Greek Inscription website enables the user to check quickly whether the form is reconstructed.
Again we hope to improve our database in the coming years, but would be happy if you could help us with this task. We have developed a system to do this (more on this...), and would be very grateful for your cooperation.