Celebrating Women’s Voices: Exploring the VOICES Project Knowledge Graph

By Lucy McKenna and Jane Ohlmeyer

Much of the historical data for VOICES – our ‘digital windfall’ – is unstructured and exists in ‘digital silos’ or what one of the founders of the WWW described as a ‘data wilderness’. So the challenge is to turn this ‘data wilderness’ into knowledge that is interoperable. VOICES addresses this by creating a ‘Knowledge Graph’ (KG), which exploits the power of the Semantic Web and Linked Open Data to answer the research questions that underpin VOICES. In the field of computer science, KG technologies are increasingly proving their worth but have rarely been adapted for the challenge posed by ‘messy’ historical data like ours.

VOICES_KG_VRTI — VOICES Knowledge Graph Demo

The VOICES KG builds on the foundational infrastructure developed by the Virtual Record Treasury of Ireland (VRTI) project. The VRTI KG for Irish History currently models over 10,000 notable Irish historical figures, primarily men, from the Dictionary of Irish Biography (DIB) as well as a present-day geo-stack of Ireland. The power of the VOICES KG is enhanced because it sits within the VRTI KG, which allows us to interlink to geographical entities and entries from the DIB already contained in the graph (many of whom are the husbands and sons of women listed in the VOICES KG). We can also access and link to digitised records (e.g. PROB 11 wills) held within the treasury itself. For the VOICES KG, we have so far focussed on adding women from four archives to the KG: The 1641 Depositions, records of the Dublin Statute, records of the freewomen of Dublin, and a selection of wills from The National Archives (TNA) UK, PROB 11. This resulted in the creation of 2,345 KG entries for ordinary Irish women from the early modern period. Prior to incorporating the VOICES KG into the VRTI KG, there were only 54 entries for early modern Irish women in the KG.

The effectiveness of the KG, of course, depends on the quality of the data that underpins it. Prior to uplift to our Knowledge Graph, the data is curated according to a bespoke person schema developed for VRTI and VOICES in collaboration with historians. The schema contains 53 fields grouped into categories such as:

Names and aliases (modernised and variant spellings)
Biographical dates and places
Familial relationships
Occupations and social status
Links to external identifiers and references (e.g., Wikidata, DIB)

The minimum requirements for an entry in the person schema is a last name, an address and dating information (i.e. floruit, date of birth, date of death), however, most entries contain much more detailed information. We capture a woman’s maiden name (where known) as her headline name, and her married name(s) as an alias. Names are modernised but variant spellings are also captured. The address is captured using a URI (uniform resource identifier) from the present-day geographical dataset in the VRTI KG. The data will, in time, be linked to an early-modern geographical dataset currently in development. Data on male relatives, primarily husbands, is also captured using the person schema but these are typically stub entries that meet the schema’s minimum requirements. Where possible, there is also a link to the original historical sources and transcripts. We hope that we can connect, through the KG, with other relevant digital projects.

Data on events and activities that women participated in and experienced is also being captured through a separate event schema. Events currently being modelled include depositions (from the 1641 Depositions), loans (from the Statute Staple records), probate (the Prob 11 Wills data) and legal proceedings (from the Irish Chancery records). The event schema captures, at minimum, the type of event in question, the dates associated with the event, the place in which an event took place, and the people/groups involved and their respective roles. More detailed information on events is captured by extending the schema for each event type. For example, in the case of the 1641 Depositions we developed, alongside our ‘person schema’, an extensive ‘deposition schema’ to reflect satisfactorily the complexity and upheaval represented in the original documents. Unique data categories were needed to capture geographical movement, widowhood and remarriage, the loss of property and the nature of violence.

The KG is grounded in the CIDOC Conceptual Reference Model (CIDOC-CRM). This is a high-level, event-centric ontology used to represent relationships found in historical and cultural heritage data, enabling the structured description of people, events, objects, and the contexts that connect them. VOICES extends CIDOC-CRM by introducing concepts specifically for modelling women, their roles in early modern Ireland and their experiences/activities during this period. These concepts form part of a series of controlled vocabularies that can be used to complete elements of the person and event schemas e.g. named titles, occupation, social status, event type, event roles, types of violence and women’s roles. Where possible, pre-existing controlled vocabularies are being used and extended for VOICES (e.g. occupational data encoded using the system developed by the Cambridge Group for the History of Population). Alongside these vocabularies sit, what we call ‘word bags’ i.e. collections of early modern words, phrases, person and place names.

Do take a moment to check out the video of the VOICES KG which features the stories of four women: Catherine Strong, Ellen Machett, Madalene Ring and Esther Van Homrigh. In the 1630s Catherine Strong, a widow who lived in the Cornmarket held the office of Dublin city ‘scavenger’. Catherine was entitled to collect tolls in return for clearing ‘great heaps of dung’ from the main thoroughfares and market areas. She invested the considerable profits she made from refuse collection by operating as a money lender, recording her loans on the Dublin Statute Staple.

Ellen Matchett and her family lived in County Armagh. During the early weeks of the 1641 rising Ellen and her family had fled their home for the safety of Dublin. En route the insurgents stripped and assaulted the convoy of refugees, wounding her mother, who rather than risking everyone’s lives, persuaded the rest of the family to leave her behind by the side of the road. It was to Ellen’s ‘unspeakable grief’ that she left her ‘wounded bleeding mother there where she dyed’. Ellen account of her mother’s death, which survives amongst the 1641 Depositions, gives voice to the extreme distress she had lived through.

Love for a mother is what stands out in a simple will made in 1623 by Madalene Ring, a 13-year girl from Waterford. Madalene was sick in body but not in mind and bequeathed all that she had to her dear mother whom she said ‘had most need of it’. A will, often written as a person faced death, might be the only evidence that a woman ever existed.

More is known about Esther Van Homrigh (1688–1723), who even has an entry in the Dictionary of Irish Biography. The daughter of a wealthy Dublin merchant of Dutch extraction, she is recorded as a free woman or citizen of Dublin, which suggests that she might be been involved in the family business. Later Esther (as ‘Vanessa’) became the muse of Jonathan Swift with whom she had a complicated relationship. She never married.

Over the coming weeks we will add to the VOICES knowledge graph 2,345 more ordinary women from four of our data sets (the 1641 Depositions, the Statute Staple records, Dublin aldermen, and wills by women that are extant in TNA, PROB 11). An initial release of the VOICES knowledge graph, easily explorable using any web browser, will be made in July 2026. Watch this space and check out our website.