On Friday April 5th I was lucky enough to attend the ISKO event Metadata, taxonomies and retrieval, which was held at City, University of London after the ISKO AGM. After introductions by ISKO chair David Haynes, 4 very different speakers covered a range of topics, all around the theme of metadata, taxonomies and retrieval.
Getaneh Alemu opened with a detailed look at how he and the library team at Solent University have improved discovery across the library catalogue, using his four main principles of quality metadata. Building on his PhD work, Getaneh outlined the four overarching metatdata principles for successful search: metadata enriching, linking, openness and filtering. Using these principles as guides, he has been able to analyse user search patterns and inform more efficient cataloguing practices at Solent.
Next we heard from Cancer Research UK’s Taxonomy Manager, Thomas Alexander. Thomas has led the development of a system-agnostic single taxonomy at Cancer Research UK, creating a common language across the whole organisation. This single taxonomy is shared across their external website, internal SharePoint site, and IT helpdesk services. While this common vocabulary has made finding information easier across the organisation, it required several stages of user testing and validation, and is slightly restricted by Sharepoint functionality in some cases. Thomas hopes that having the solid taxonomy established will be a great place to start any future ontology projects he has in mind.
After a refreshment break, Niké Brown presented around taxonomy challenges in digital publishing. Niké has many years’ experience in this field, and even remembers her former company launching their first ever CD-Rom, a format and content challenge that helped launch her particular interest in this topic. We were taken through several examples of different approaches to digital publishing taxonomies, and the challenges faced by each. Several points were raised linking to the previous presentations, around the pros and cons of one large, all-encompassing taxonomy vs. multiple smaller, specialist taxonomies, and the importance of proper metadata enrichment.
The final speaker was Eugene Morozov, who talked about W3C Semantic Web standards and discoverability of regulatory rulebooks. Eugene went over some of the pros and cons of Model-Driven Machine Executable Regulation (MDMER), and the benefits this developing field could have in the future. Building on the 5-star Open Data rating scheme, Eugene provided examples of various regulations and rated them accordingly, including an axis of granularity and an axis of semantic richness to provide a fuller picture.
The audience was then invited to ask questions of the 4 speakers in a panel discussion, addressing the questions Metadata and Taxonomies: best hidden behind the scenes, or fully exposed to users? All speakers had different experiences of both these approaches, and all ultimately agreed it depended on context. There was a consensus around the importance of having enough information for users to be able to narrow down their search, help with ambiguous search terms find related terms etc, without providing an overwhelmingly large full taxonomy. Some interesting points were then raised around AI and machine learning, as well as voice search and how taxonomies will apply and be available for voice search results. All panellists believed the most important thing regardless of the approach adopted is having high quality taxonomies in place.
Overall it was a thoroughly interesting afternoon, with some great speakers and thought-provoking questions. I look forward to the next ISKO event! https://www.iskouk.org/events