Wikidata
2023-10-26
If you execute the above query, you will get in alphabetic order and with English (Latin) names all entities in Wikidata that are mentioned in Pausanias’ Description of Greece.
If you write "el," before [AUTO_LANGUAGE], or just "el", removing all other languages, you will get the same result with Greek Names.
In both cases, you can download and save the query results as CSV Files. The following two tables are created from these CSV Files and display the query results in English and Greek.
Pausanias mentions approximately 3000 real or legendary human beings or divine characters, often several times in widely dispersed sections of his text. The need for unambiguous named entity identifiers was self-evident. Rather than create our own (as Manto does for mythical characters) we embraced the opportunity to use the preexisting IDs offered by Wikidata in our person annotations. This also allowed us to dispense with the work of tagging known persons with conventional (but ontologically problematic) tags such as historical/mythical or male/female, since such properties are almost always present in Wikidata items.
In 2020, when we began the annotation process, automated alignment to Wikidata using (for example) the Wikidata lookup feature in OpenRefine was not a realistic option. Though ultimately we proved able to find an existing Wikidata item for 85 percent of the characters Pausanias mentions, many of those items had labels in languages such as Catalan or Slovak with unfamiliar orthography, missing or useless descriptions (“mythological character”), or too few data statements to disambiguate ancient names that might refer to three or thirty different individuals. We therefore decided not to generate and upload to our Recogito annotation platform a Wikidata look-up table for persons analogous to the Pleiades/ToposText/Arachne gazetteers we adapted for places and artwork. In retrospect, this decision cost our Wikidata specialist time and labor, since a high percentage of Pausanias’ persons are reasonably identifiable.
In their first pass through the text, Periegesis annotators simply marked each instance of a person using the Person type in Recogito. Once this pass was complete, the annotations were exported as csv and Persons were pulled out as a separate table and compared to an equivalent persons table, incorporating patronymics where available, from the English text of Pausanias in ToposText. Meanwhile, that ToposText keyword-in-context index, with its idiosyncratic entity tags, was laborious aligned to Wikidata manually over a couple of months via the Wikidata search box, using various languages, including Greek, and diving into the underlying Wikipedia articles when identity was unclear. English labels were added or corrected where necessary, and basic descriptions and other data statements added to make items more findable. A simple script concatenated paragraph ID with the target entity to substitute the appropriate Wikidata ID for the ToposText tag in the English text of Pausanias. A similar script, using normalized transcription of the Greek, aligned Greek person entities to English ones.
Once this process was mostly complete, each Recogito annotator was given an ordered list of persons by paragraph with Wikidata IDs the annotator would copy-paste as a tag in the existing Person annotation in Recogito. When no Wikidata item could be found, Persons were assigned a “WD” tag to locate them, plus gender and other tags, and the Comment panel was used to create an English label for a future Wikidata item that could be created in QuickStatements or manually.
The net gain from this arduous process is a uniquely complete named entity annotation of a complex ancient text with Wikidata identifiers, plus a substantially enlarged and improved body of Wikidata identifiers that can be uploaded into Recogito and other annotation platforms to perform semi or fully-automated tagging of persons from classical Greek civilization.
The following query in the Wikidata Query Service will generate a list of Wikidata entities (currently 1900+ but it will expand) marked as having been described in Pausanias’s Description of Greece, i.e., using the property P1343 "Described by source" and the object "Description of Greece (Q3825645) work by the 2nd century CE traveler Pausanias"
The attached csv file pulls up Wikidata identities with English descriptions and properties for approximately 21450 persons and religious terms from Greek and Roman antiquity, a list that can expedite assigning named entity identifiers to texts dealing with Mediterranean mythology and history.