Sixth ISKO UK Biennial Conference, 15-16 July 2019, London

by Monique Archer, University of the West of England, Bristol

It was a privilege to attend the sixth ISKO UK Biennial Conference, the Human Position in an Artificial World: Creativity, Ethics and AI in Knowledge Organization. The conference was hosted by City, University of London. I am grateful for a very informative and thought provoking two days and the opportunity to network and meet fellow colleagues in the field, exchange of ideas and to listen to the presentation of my lecturer Dr Paul Matthews.

The conference covered many topics about AI, Ethics, Creativity and the human position. Here are a few highlights:

Ludi Price, a visiting research fellow at CityLIS and one of the winners of the best paper awards spoke about her research into "Fandom, folksonomies and creativity: the case of the Archive of Our Own (AO3)". Ludi’s presentation was based on her doctoral thesis and her aim was to ascertain three things:

1. Tag analysis
2. User's intent
3. Preservation of original tags by "tag wranglers"

The methodological steps of the research aligned with my own dissertation. I thoroughly enjoyed this presentation which led to a follow-up discussion with Ludi, which, in turn, helped with the interview aspect of my dissertation. The presentation took you from democratic indexing to fandom, to different types of fandom. Ludi explained how the tag analysis was performed using degree and betweenness centrality, clustering and density. In the end it was an example of the dynamic nature of fandoms and how persons within such a realm view the work they do to maintain such folksonomies.

The other best paper award went to Shu-Jiun Chen, from Academia Sinica, Taiwan, who spoke about "Semantic Enrichment of Linked Personal Authority Data: a case study of elites in late Imperial China". She showed how linked data can be used to enrich data about individuals mentioned in Chinese archives from the late Imperial era. Some of the techniques used included instance-based semantic enrichment, linking data to other sources, adding missing data to the sources and applying time ontology to the knowledge base.

The paper by Valdir Amancio Pereira Junior, Gustave Pereira and Leonardo Castro Botega was entitled "Towards a Process for Criminal Semantic Information Fusion to Obtain Situational Projections". The presenter explained how situation awareness and semantic information fusion were applied to crime reporting to improve the quality and reliability of the resulting reports. Adding a layer of semantics and ontology application further helps in the process of information analysis.

Tanja Svarre and Marianne Lykke of Aalborg University, Denmark spoke about "The Role of Knowledge Organizing Systems in Business Intelligence: a literature review". They noted that metadata and ontologies played a key role in the user experience of KOS in the usage of Business Intelligence systems.

The IKO Case Study Café was an interactive event where we had the chance to engage in in-depth discussions with a selection of case presenters. This session was the great highlight for me at this conference. Eight case studies were presented to the audience with the invitation to select three for discussion with the presenters. The three case studies I chose were:

1. Using Knowledge Graphs to Model Standards and Business Processes in the Testing, Inspection and Certification Industry - Ian Davis, SGS
2. Using Distributed Ledgers (AKA Blockchain) to Enable Trusted Exchange of Commercially Valuable Information across a Defence Consortium - Marcus Ralphs, Byzgen Limited.
3. Beyond Posting Counts: Giving Taxonomists a 360 Degree View of How Concepts are Being Applied to Content - Dave Clarke, Synaptica.

The second day of the conference began with the keynote paper on "Creativity and AI" presented by Neil Maiden of Cass Business School. The question here was: what forms of AI capability can augment human creativity in different professional environments. Based on his interpretation of creativity as information processing, the author reported on a series of experiments carried out in the context of health and safety in manufacturing, journalism and the creation of personas. As far as creating search strategies, landscaping analysis, generating ideas and challenging constraints, it was clear that human interpretation and judgement was still needed.

Vanda Broughton (University College London) dealt with the issue of bias in the process of intellectual creativity and automation in representing diversity, in this case religious diversity. The hope is where the AI may fail human intervention can rectify such biases.

Dave Clarke of Synaptica completed this session by demonstrating how we can work in the space between taxonomies and ontologies, the latter being more useful to AI, thanks to their structure which reflects reality better.

The following session on "Creativity in AI" continued with David Haynes, a research fellow at City LIS, who spoke about creating an ontology of risk, why it should be created? How it can show relationships between concepts and to map these risks to help reduce risk and offer solutions pre-emptively?

This led nicely into Paul Matthews' (University of the West of England) presentation "Human-in-the-loop topic modelling". He suggested that the human element is critical for any AI optimal functionality to be successful. He used a movie plot summary corpus and topic labelling as the example how machine learning can help to interpret relevant data to produce results.

This session was followed by a lively panel discussion on "In an AI supported world, where do opportunities lie for knowledge organisation?" One key take away is almost everyone agreed that the human aspect must be present in knowledge organization, particularly for ethics and dealing with bias.

The final session of the conference rounded off with papers dealing with searching and AI and the benefits and challenges these can bring to bear on KO systems. Where information needs are placed at the centre of each research and to apply AI suitably for the best experience possible and this can be done by reducing syntax errors and query expansion to name a few.

The conference also included posters. Two that I found particularly interesting were:

Mike Kelly, University College London - Visualising knowledge in Cultural Heritage former topic speaks to ways in which cultural heritage can use a visual vocabulary to show meaning and illustrate heritage research and some of the issues relating to diffuse datasets and interpretation.

Linda C Smith, University of Illinois at Urbana-Champaign, USA - Artificial Intelligence in information retrieval: forty years on gave a timeline of forty years of AI in information retrieval. The author was a witness to AI research from 1976 to the present and was able to provide a personal account as well as a scholarly perspective informed by the literature.