By Katherine, Amara, and James (Project Investigators)
Constructing an ontology around relations between people is (relatively) straightforward when those relations are formal or familial: person X was taught by person Y, person A shared a house with person B, person P was the daughter of person Q. Where it gets difficult is when those relations are not familial, are less tangible, and we feel those relations are both significant and not unambiguously captured in our sources: for example, when two or more people appear to have been friends and that friendship appears to have been important to their lives.
Wikidata does not have a property for friendship, and as a project team we decided – after lengthy discussion and disagreement – that creating a property for friendship was unwise, because ‘friend’ is highly ambiguous. For example, ‘friends’ is used in ‘friends, Romans, countrymen’ to hail a group of people who are not necessarily friends, but rather to encourage those people to treat the speaker – or the institution they represent – as if friends. Similarly, the expression ‘friends of the British Museum’ is an organisational usage that describes a loose group of patrons, donors, and/or sponsors who are not necessarily ‘friends’ with the organisation but are rather well-wishers and/or supporters.
Wikidata – which Beyond Notability treats as a guide, though something short of a model – gets around this issue with the ‘significant person’ property, defined as ‘person linked to the item in any possible way’. That is, the property is used to define a range of relations that are significant to the subject: for example, between pi and William Jones (the latter named the former), Albrecht Dürer and Willibald Pirckheimer (the pair were noted correspondents), Hispano-Suiza and Marc Birkigt (the latter founded the former), 284 AD and Diocletian (the latter became Roman Emperor in 284 AD), Sarah Bernhardt and Betty Callish (they were friends).
For us, adopting this approach would have three disadvantages. First, it is non-specific, and a key rationale for creating our own knowledge base – rather than simply augmenting Wikidata – is to allow us to drill into the specificity of relations that our domain demands. Second, it is implemented in Wikidata without the requirement for evidence (to get technical, there is no property constraint for P3342 that requires a valid reference). Third, and more philosophical, is the emphasis on ‘notability’ (both of the subjects involved and of the relationships between those subjects), and the positional work that implies – notability thresholds depend on whosenotability counts and which relationships are considered to be significant. In a project seeking explicitly (it is in the title!) to go beyond notability, we have already resolved to always to make explicit that we are ascribing to people aspects of identity and self-image perceived by us in our interpretation of our sources. In turn, constructing our own notability thresholds for things like friendships between people – for, per Homosaurus “Connections based on affection and trust between two or more people” (and the various cascading private and public specificities thereof) – struck us, after much debate, as unwise.
So, what ascriptions or expressions of connections of this ambiguous kind are we encountering in our sources?
We know that Maria Millington Lathbury and Jane Harrison lived for some time at the same address – Chenies Street Chambers, a women’s residence in London. They were interviewed together in Pall Mall Gazette about lecturing work, and Harrison was described by Lathbury’s daughter in a published memoir as one of the “friends” who sought to help her establish her career as a freelance lecturer.
We know that Agnes Conway dedicated her 1917 book to Jane Harrison describing her (in Latin) as a “beloved teacher” (“magistra dilecta”); and that Conway’s diaries reveal Jane Harrison was a hugely influential mentor for her for many years after she left Newnham College. For Amara, who has extensively researched Conway’s diaries, this amounts to friendship.
And we know that Anna Anderson Morton and Mary Brodrick ran a business together, travelled together, and were likely to have shared an address for part of their lives. Morton also arranged public appearances for Brodrick, announcements for which appeared in contemporary newspapers.
Based on our discussions, ontological observations, and analysis of evidence, our solution – for now – is to create the property ‘has personal connection to’. We plan to use this property to capture instances where evidence exists of non-familial and personal connections between people (beyond merely using the words such as ‘friend’ in a formal manner) and that is a more modest way of asserting connections between people that evoking ‘friendship’. For example, we have used this property to capture the personal connection between Ethel Henrietta Rudkin and Margaret Alice Murray as evidenced by Murray writing a foreword to a book written by Rudkin and by Murray having been noted as encouraging Rudkin’s work. This usage is then similar to Wikidata’s usage of ‘significant person’ in that its use is likely to be sparing – we don’t anticipate that every women in our knowledge base will have a ‘personal connection’ statement associated with them. However, it differs in placing an emphasis on the personal, and in not picking out ‘significant’ relations but rather trying to amplify the significance of other statements that suggest personal connections – in making ‘personal’ the evidence that links flatmates, colleagues in an excavation, or committee members.