The KMWorld 2017 Conference – including the co-located conferences Enterprise Search & Discovery, SharePoint Symposium, Taxonomy Boot Camp, and Text Analytics Forum – took place November 6-9 in Washington, D.C. Although I didn’t personally attend every sub-conference or session, I did have enough conversations to identify some common themes.
Probably the most significant buzz, hype, and confusion centered around the nature and role of artificial intelligence and machine learning. While true AI might be longer in arriving, machine learning can be applied today within the enterprise. Successful use of ML often depends on the industry and the use case. Some problems are more conducive to machine learning techniques than others. A common feeling among participants was machine learning is a developing technology and the promises don’t always match up to reality.
The most prevalent and surprising reaction to ML was the acknowledgement that it is an immature technology and practice in the enterprise. As such, there is a significant amount of work and planning required to launch a successful organizational practice. More often than not, enterprises want to embrace and launch new technologies to get ahead of their competitors. Thus, they are willing to spend money to shorten the length of time it takes to get up and running on new tools or with new practices. However, a trend I noticed was the understanding that machine learning was not a simple proposition to embed in the enterprise rapidly. Most business and technical people are willing to spend the time and effort to get machine learning right. They know that a failed launch is far more difficult to recover from than a delayed launch.
For many business users, and even some technology representatives, a deep understanding of machine learning is elusive. It requires external expertise and technologies. Reliance on external resources isn’t at all surprising as any shift in information practices and technologies requires outside expertise to get started. I think there is a willingness on behalf of organizations to lean on external vendors and consultants to help guide them through the murky, difficult world of machine learning.
Taxonomies & Ontologies in the Enterprise
Another surprising, yet satisfying, observation I made was the number of people who were knowledgeable about taxonomies and ontologies, their roles in a greater information landscape, and what difficult problems they believed the use of controlled vocabularies could solve. There are always newbies and novices at KMWorld, which demonstrates not only a continued relevance in the field of information management but also a willingness to dive in and learn about an area which may not be familiar. While this was still true, I was astounded at the level of expertise.
I was asked the usual question, “What is taxonomy?” fewer times this year than any year I’ve attended in the past. If we look to Gartner’s Hype Cycle charts, perhaps this shouldn’t be so surprising as taxonomies and ontologies are reaching maturity and productive use. The satisfying aspect is not just the excitement of getting to more advanced practices, but knowing that years of education and successful enterprise projects are coming to fruition.
Text Analytics Forum
In related news, it was the inaugural conference for Text Analytics Forum and the sessions were well attended. While text analytics as a topic has shown up in many sessions in the past, it has never had a dedicated conference. The fact the subject now has its own event and generated a lot of interest speaks to the importance of text analytics and the recognition of its place in the enterprise.
Text analytics is one foundational activity in search, taxonomy and ontology building, and as a tool for analyzing the growing body of semi- and unstructured text in organizations. Natural language processing and the ability to process and analyze text is a component of machine learning and artificial intelligence. The interrelatedness of the disciplines was apparent in the session topics and attendees’ areas of discovery. Interest in text analytics as a discipline in and of itself was probably matched by interest in it as a component of more advanced machine learning applications.
While the usual themes of information governance, retrieval, and collaboration were present in the discourse, the expectations of end users seemed higher than they have ever been. It is an exciting time to be in any of the fields related to managing and using information, and the excitement was palpable at KMWorld.