KMWorld 2024 Is Nov. 18-21 in Washington, DC. Register now for Super Early Bird Savings!

Adding Cognitive Features to KM Part 2 (Video)

Video produced by Steve Nathans-Kelly

Nalco Champion's Dynisha Klugh and TechnipFMC's Kim Glover discussed how and why predictive search autosuggest, chatbots, and other cogntitive features are useful during a session at KMWorld 2019. 

Search autosuggest is targeted at users who would like predictive help when they are serarching, said Klugh. "And we all see this when we use the internet--it's kind of the type-ahead: we're typing and if I say, 'where can I,' there are five different sentences already in the drop-down that I can select from. So, that is predictive auto-suggesting of things. And, again, it is based on personalization for me, so the system knows where I am and it knows the region I'm in, it knows the other kinds of things I've looked at in the knowledge environment. So when I'm searching, it's helping me to get smarter and maybe even think about something that I wasn't aware of so that I can get content before I have an issue."

For a while, said Glover, her company had Smartlogic from Semaphore, and it also did auto-classification. "It is really wonderful, if you can have a tool that does that, I would highly recommend it. It still needs the user intervention sometimes because it doesn't always look properly into the document, or maybe the document can sort of seem to be about something else, but it can take you leaps and bounds ahead in terms of findability."

It can allow a user to upload a file, said Klugh, and the machine will read through it, proposing tags. Human intervention is somebody refining that, maybe, adding synonyms to it, some things unique to just your company that you might also want to put in there, but not trying to have a full taxonomy tree that you have to maintain, but more of a living classification system because the machine is getting smarter as new documents are uploaded into it.

In TechnipFMC's case, said Glover, "We actually did have a 9,000-word taxonomy, but they worked together very well and I still wish for those old halcyon days."

To retrieve content, said Klugh, "our mantra around our discussion board is search first, then ask. And the reason isthat you're going to get a faster response because your answer might already be sitting there because somebody had a similar question a month ago and you didn't know it. So before somebody asks a difficult question we say, go and search. And our users are saying, 'As a user I want to be able to search and have a bot help me out.' So, I search--you think of a chatbot, like we all see when we go to a shopping site, the little thing is sitting there saying 'How can I help you?' You click on it and you can search with that. Not only will it retrieve content, you can rate how useful it was. So the bot can say, 'Here's what I found on product EC215A, was this helpful? The bot is getting smarter and refining its results based on user and crowdsource responses in terms of how you rate it."

Video produced by Steve Nathans-Kelly

Videos of KMWorld 2019 awards presentations, keynotes, and many sessions can be found here.

Many speakers at KMWorld 2019 have also made their presentations available at www.kmworld.com/Conference/2019/Presentations.aspx.


KMWorld Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues