A week on from the ISKO UK 2017 conference: Knowledge Organisation – what’s the story? (#ISKOUK2017) I’ve finally had time to gather some (actually quite a few) thoughts on two days mixing knowledge organisation systems and narrative. It’s a full and thought provoking story full of fake news, future shock, linked open data and ingenious knowledge organisation systems.
These are my thoughts in response to the conference theme and presentations but where they refer more directly to a conference paper, or an external reference, a hyperlink is provided so you can go and seek out the original inspiration.
All the conference papers and recorded audio will all be available on the ISKO UK website shortly.
Stories Matter
There are stories as well as structures in our knowledge environments and we need to think about the role of knowledge organisation in storytelling.
Stories can be effective at transmitting knowledge, especially in less structured environments to such as liminal spaces/liminal moments.
Stories are useful for information sharing between coworkers and shaping organisational culture.
Stories enhance research in digital archives by adding contextual relationships to inventories. Narratives in archives can be revealed through the ‘conscious intervention’ of archivists.
Stories are not necessarily antithetical to the facts (make believe); there’s a truth in the synthesis of models and evidence that can sometimes best be revealed through narrative (make sense).
But the plasticity and volatility of stories makes them hard to organise and use to convey knowledge systematically and reliably.
The semantic web is the (five) star
There’s was a lot of enthusiasm for RDF, graph databases and the potential of linked open data to connect concepts and stories across domains.
There were many examples combining KOS with predictive analytics to expose unobserved narratives in unstructured data.
In research, PLOS are using their taxonomy to filter out fake science, speed up peer review and analyse manuscript submission narratives.
Certainly open research would benefit from linked open data and a more semantic research graph to analyse and visualise new research narratives through more meaningful interfaces such as the ‘Tinder-like’ research discovery prototype Aalborg university have created to share research with SMEs.
We can also think of research questions, not just outputs, as research metadata and use knowledge organisation to make them smarter.
There’s an possible vision emerging of 5* data for 5* research: semantically rich, linked open data for research intelligence.
In healthcare, the application of ontology to data silos can extract facts that when compiled in a graph database predicts and reduces incidents of trips, slips and falls.
In broadcasting, the use of automated data architecture based on Wikipedia categories via linked open data has successfully opened up radio programme archives to topic based exploration with minimal editorial effort.
In information retrieval, linked open data can help search return answers not content by understanding questions and filling in the semantic gaps.
For automation, concepts in knowledge organisation systems need to be machine readable, unambiguous and uniquely identifiable.
There were issues: for example, concepts don’t necessarily cross domains meaningfully.
In many cases, linked open data eventually connects back to a few key ‘anchor’ datasets like DBpedia or Wikipedia. It concerns me that implicit bias in this anchor data potentially structures multiple knowledge organisation systems.
Are facts sacred when comment is free?*
Day 2 aired concerns the proliferation of fake news and false narratives in social media and personalised search is both disrupting knowledge and distorting democracy.
Post-truth was the Oxford Dictionaries word of the year in 2016, reflecting an explosion of content that shapes public opinion based on appeals to emotion and personal belief rather than objective facts.
There was much thought provoking discussion exploring the trends contributing to this phenomenon and what the response of the knowledge organisation community should be.
Social media collapses boundaries. It decontextualises content and doesn’t clearly delineate genres. As attention has migrated from authoritative sources towards social media information has become more open but more blurred: it is becoming harder to separate fact and opinion.
But then, all recorded knowledge is fundamentally decontextualised and not intrinsically bound to the narrative that produced it. It can be restructured and reorganised into other narratives.
Consumers love personalised search and its application has spread from retail to information retrieval. Homophily allows people to recognise rather than memorise when navigating information and easily find things that appeal to them.
What organisation of knowledge exists on social media is hidden within opaque algorithms that tailor and target content to suit personal preferences and ad dollars.
Social media as a technology is situated in wider trends of individualism in a divided society that privileges subjectivity. We seek and share information that reinforces our views and undermines those we disagree with.
Social media is real time. Now the first draft of history is often written rapidly and out loud. Factual errors have become lore by the time the verified record can correct them.
It turns out social media is an unreliable narrator: at times truthful, at others half-baked, inaccurate or worse: deliberately and covertly mendacious.
Social media is noisy. Despite some limited content curation features, publication is often impulsive, spontaneous, repetitive, un-selective and eschews editorial.
As data proliferates and accelerates it outstrips the capacity of literacy and journalistic methods to separate signal from noise; so we too frequently delegate to this task to trusted grapevines or algorithms.
Social media favours anonymity over authority. Few accounts are from verified identities, the only requirement is an email address. Most identities are self-asserted, many are totally made up.
None of this is in and of itself a problem.
Social media is problematic when it is the only source people use for all their information and news to the exclusion of any other genre or source.
Social media is problematic when filters exert influence at a subliminal level and often exist without consent, control or awareness. This leads to the filter bubble effect: a distorted and individual view of reality.
It is an issue that people are unaware the content they see is increasingly personalised and partial and reflects their profile rather than the world as it is.
Social media is problematic when people are unable or unwilling to distinguish fact from fiction and give all content and sources equal weight and credence without thought.
Social media is problematic when entertaining is synonymous with informative. When the performative, the subjective, the diverting, the emotional and the recognisable attract and hold too much of our attention over the critical but sometimes tedious work of verification and rational thought.
We might not want to be lied to, but the paths formed by our digital footprints tell the story that misinformation is profitable and the cost of accuracy a price were are not prepared to pay.
Social media is problematic when used for dark politics or propagating hate. Advertisers buying into thousands of data points use extreme demographic profiling not just for marketing but for political messaging and spreading propaganda.
If we use algorithms and machine learning to create classification schemes will they generate a map of knowledge or bias? At what point is targeting unethical, even illegal?
Truth seems too philosophical, lies too political. We are really dealing with the intermingling of facts (verified information), rumour (circulating propositions) and lore (accumulated belief) and our tendency to mistake one for the other.
The neatly landscaped borders of knowledge are increasingly wild and overgrown and we are not good farmers or foragers.
Facts Matter
This is a crisis of verification, stressing existing methods of assembling, checking for accuracy, bias and relevance and consuming current events (journalism) and evidence (scientific research).
If we are to restore truth to power we need first to reclaim trust in expertise and agreed methods of asserting authority.
Facts Matter is a campaign run by CILIP that aims to reassert the central role of evidence, verification and authenticity in public life.
This is a huge editorial challenge for journalism, for publishing and yes for technology platforms who are not simply data carriers but are knowledge organisation systems and conscious interventionists too.
Impartiality is not the appearance of neutrality: it’s transparency, traceability and being as close to the truth as possible.
It is also a huge literacy challenge for knowledge professionals, libraries and educators to equip citizens with accessible, authoritative information and the skills and critical faculties they need to navigate hypermedia.
To help us interrogate massive datasets truthfully we need to consider the provenance and purpose not just of sources but also structuring knowledge organisation systems, platforms, filters and funding.
In doing so we must also be careful that any solutions to these challenges don’t introduce other problems or unintended consequences; many also rely on data mining and automation.
We must also guard against holding too closely, too uncritically the things we recognise and hold dear and accusing of being false or ignorant things we simply don’t agree with.
We must take care not to infer too much about the influence and impact of fake news from the limited evidence we have as yet about its proliferation.
We must define and agree what we mean when we speak of misinformation, disinformation, fake news and false narrative and map out the knowledge ontology of hypermedia.
Evidence is always necessary but insufficient; we also need the application of research methods to formulate and test hypotheses, to synthesise evidence into firmest known facts shared through engaging narratives.
Underlying all this is a quest to surface bias: whether explicit (to further an agenda), implicit (built into systems) or tacit (deep, unconscious).
If it’s not possible for any system to be completely unbiased then any bias, the propositions on which it is based, should be conscious, overt and open to debate.
KOS are narratives themselves: they are accounts or stories of how knowledge is organised. They are part of the epistemic communities in which they function and perpetuate the biases of those communities. They are inserted in broader narratives.
We are not beyond truth or post-truth. Understanding truth remains an intricate philosophical questions of our time as well as a pressing social issue, policy debate and technology challenge. It demands further investigation, careful thought and innovative solutions that uphold the principles of both free speech and the common good.
The knowledge story continues …
So, stories and facts both matter. We need to design systems to organise the knowledge they both encapsulate that are accessible and transparent in their bias so people can make informed judgements and choices.
We we need new kinds of literacy, we need more diversity, we need flagging, we need checking and debunking, we need not just open data but open sources, standards, platforms, filters and algorithms.
We need to rethink genres and devise ways of clearly signalling genre, topic, source, provenance, authority and certainty in new media. We need to revisit assumptions and and ways of thinking not just more technology.
We panic when there’s too much too know, when our knowledge organisation systems are overwhelmed, technology changes, when established paradigms are challenged.
Dominant paradigms are crumbling, or being deliberately weakened; emerging paradigms are immature and untrustworthy; residual paradigms are beloved but nostalgic. This is the unstable terrain of any culture shock or paradigm shift.
For now it’s social media, but soon mainstream virtual reality or advanced robotics may come and change narrative paradigms and our experience of fact and fiction once again.
Still, the historical narrative of KO tells us that from Llull’s Ars Magna, Harrisons’ arca studiorum or Otlet’s Mundaneum to contemporary hypertext, algorithms, artificial intelligence and beyond we continue to model, invent and evolve thinking machines and ways of mapping knowledge that cope with the shock of the new.
- “Comment is free, but facts are sacred” is a famous sentence from Guardian editor CP Scott’s essay ‘A Hundred Years’ written in 1921 to celebrate the centenary of the Guardian newspaper. The full paragraph shows that issues of truth, propaganda and fairness were preoccupations of that age too:
Neither in what it gives, nor in what it does not give, nor in the mode of presentation must the unclouded face of truth suffer wrong. Comment is free, but facts are sacred. “Propaganda”, so called, by this means is hateful. The voice of opponents no less than that of friends has a right to be heard. Comment also is justly subject to a self-imposed restraint. It is well to be frank; it is even better to be fair. This is an ideal. – CP Scott, ‘A Hundred Years’
16Blue 