The annual KMWorld Conference went virtual this year as KMWorld Connect from November 16th – 19th with the pre-conference workshops running November 12th–13th. The co-located conferences, Enterprise Search & Discovery, Taxonomy Boot Camp, and Text Analytics Forum were included, but Office 365 Symposium was cancelled.
The geographically distributed Synaptica team was “there” at the conference representing our products, giving presentations, and attending sessions. Distributed across multiple time zones, the team got up early, stayed up late, and did our best to be present at the conference based in U.S. Eastern time.
The Virtual World
Moving one of the largest knowledge management conferences online is no easy task. We were able to meet via the PheedLoop platform, allowing us to be present at our virtual vendor booth to chat and answer attendee questions and attend sessions from the four conferences. One advantage to attending a virtual conference is the ability to be in multiple locations at the same time. While attending sessions, it was also possible to answer chats and emails as they came through. Likewise, it was possible to see audience questions in real-time during sessions and, as a presenter, prepare to answer these questions at the end of the presentation.
Of course, the point of conferences is to meet known colleagues and new people face-to-face, and, despite the ability to chat and share video, this was simply not the same in the virtual world. Similarly, the after-conference events which often spontaneously occur as people form groups to grab dinner and engage in additional discussion were noticeably absent.
Despite the shortcomings of an online conference, the quality of the content was spectacular as usual. Let’s break down some themes.
Knowledge Graphs, Knowledge Graphs, EKGs
If you didn’t hear the phrase “knowledge graph” at least once, then you didn’t attend KMWorld. The new associated acronym is EKG (enterprise knowledge graph not electrocardiogram), which is certainly appropriate for measuring the excitement around knowledge graphs.
We have discussed our viewpoint on knowledge graphs in our blogs, webinars, and workshops, and, in short, we believe that a knowledge graph is an ontology defining the properties and relationships of a scheme, the taxonomy values which populate that scheme, and the connections to information sources. Despite some variances in definitions at the conference, knowledge graphs have earned their place in the discussion for several reasons.
First, in my opinion, knowledge graphs are popular across disciplines for the very simple reason that they are intelligible. Ever since the birth of the semantic web, we have been trying to model our world, a world we can make sense of, in the electronic universe. The foundational connections in an ontology and their implementation in graph databases as triples (subject, predicate, object) literally mirror the way in which most languages work: subject, verb, object. The connection of these subjects and objects with relationships to define the world is appealing to people in knowledge management, who value the relationships between people and people and things, and to those technical folks who must model and reason over this connected graph. We’ve seen this implemented on social media platforms like Facebook and LinkedIn, and it’s very effective because we can understand the reasoning behind it.
Second, knowledge graphs are made possible and accessible to more enterprises by advances in technology, including the ability to define graphs more easily in graph databases (while still being able to retain and access relational databases), the ability to store and access more content whether behind a firewall or in the public domain, and by changes in the way that we work as employees within the organization and remotely. Simply put, affordable, scalable technology can support active and workable graph structures.
Finally, taxonomies have reached maturity in the industry. What was once an esoteric and obscure activity has gained traction in enterprises and, with this, the need to support more complex knowledge organization systems such as ontologies. While certainly not every taxonomist is an ontologist, nor does every taxonomy need to become an ontology, the art and science has advanced and become more commonplace.
Artificial Intelligence & Machine Learning
The promises of artificial intelligence, or, more specifically, machine learning, have been at the forefront for years now, and this KMWorld was no exception. The difference that I noted this year is how machine learning fits into and supplements knowledge management rather than just being a technology which could auto-categorize and serve up content and answers.
I think this is an important shift and marks a change from machine learning as back-end and black-box technology to front end and potentially explainable knowledge management support. While KM is really about people, machine learning has the potential to aid in analyzing and even creating relationships between people and things; specifically, content and knowledge.
Machine learning models can be trained to recognize patterns in data, and these patterns, once recognized and isolated, can be interpreted by both machines and people. Pattern recognition and analysis is essential to businesses to note trends in the marketplace, future events, and even the behavioral patterns of employees which can lead to a more productive, safer, and enjoyable workplace.
Chaos & Complexity
Perhaps unsurprisingly, there was a lot of discussion around chaos and complexity. Some of this was probably indicative of the numerous, chaotic catastrophes of 2020. Equally as probable is the pertinence to knowledge management of continued attempts to manage the chaos and complexity of information through modeling in knowledge graphs and use of machine learning in data analysis.
A common theme was sense-making of complexity. In a world of ever-growing amounts of data and content, how do we make sense of it and find common meaning? How do we recognize patterns and make predictions in large, complex systems like the global economy or a global pandemic? What can we do to prevent catastrophes or respond to them more quickly in the future? In tandem with the explosion of content comes an explosion in meaning and attempts to pin meaning to information in which various contexts may imply various meanings. In one context, the meaning is clear; by changing the context, the meaning is torn apart.
Chaos and complexity is relevant to knowledge management because information is created with intent, and the context can convey that intent. Knowledge unmoored and out of context can create worse problems than not having the information at all. Hence, it is up to professionals in the field to attempt to retain context and sense when creating, managing, and presenting information.
The common thread of governance is also thematically tied. While you can’t control the world, you can control the way your content is created and organized. You can control the way taxonomy terms are added to a taxonomy and how the taxonomy is used in consuming systems. You can control, or try to control, SharePoint sprawl. The point is, while the world at large is chaotic and complex, it is possible to institute processes and procedures more locally to rein in what can be controlled.
In the controlled vocabulary world, governance is nothing new. However, as models become more complex and drive more and more consuming applications, including machine learning, it becomes an imperative to control the information presented and data flowing into those systems. While governance has always been a lynchpin in ensuring content can be categorized and found, there is now more at stake. Recognizing objects in real-time by autonomous vehicles, accurately conveying information in the throes of a pandemic, or models used to predict natural disasters all rely on clean, well-governed data.
While KMWorld has often covered the same topics as the industry has progressed and changed, the real-world context forcing us to attend a virtual-world conference has brought the intertwined themes discussed above into sharp focus. We look forward to continue working on untangling the threads and weaving a semantic web for better knowledge management.