From knowledge management to knowledge discovery

I spend a lot of time dealing with data. Data is an ingredient in an organization’s larger decision making process. Institutions say they want “knowledge management” systems. This is a way to store all the information they create in one place and make it searchable. But the knowledge solutions I’ve seen only address half the problem. They focus on consolidating information into one data warehouse. They treat the people that have to search it as invisible, anonymous, and interchangeable.

I believe that this is less useful than designing systems that alert users when something interesting is available within the context of their daily workflow. This, for me is knowledge discovery through implicit inquiry. I work on 20 different projects a week. I have neither the time nor the head-space to stop and study the history of the concepts behind each one of these. I want any knowledge I need to come to me when it’s important. I do all my work in systems that leave a trail of breadcrumbs that would let a search algorithm know what I need.

For an implicit inquiry algorithm to work, you need workers to work inside a system, like Slack, and then it is easy to detect what people want to learn about, based on the conversations they are having at any given moment. I would combine this data with a model of the person based on what they have produced in the past (e.g. what documents they’ve authored on the company’s Dropbox account). Now, a robot could alert me when there’s some other file in Dropbox that is a strong match for what a person is talking about, and what that person has worked on in the past, but that lives outside the files and people that person has recently collaborated/interacted with.

I call this knowledge discovery, and I expect it to be an intuitive and unobtrusive part how we work in the future. Context awareness is the key missing ingredient:

I’m starting to build it within the slack+dropbox+gmail knowledge universe. There’s enough context there to make it meaningful already. Adding in the feedback signals from the Feedback Commons would make it timely as well. Facebook has already done this for marketing – it knows what we like and don’t like, and what we’ll buy, and it sells ‘us’ to advertisers. Google has done this for search – it knows what you’re after each time you search in part because of what you’ve clicked on before and where you’ve been. Is Slack going to unlock this for workers in workflows in nonprofits? I’m betting so.

At Keystone Accountability, I have been gathering tools that help organizations better understand many types of information they can gather about the communities they serve. Lately I have paid closer attention to what meta data those tools can generate that might inform an implicit inquiry robot, to make robotic knowledge discovery possible.

We all want evidence to transform people’s lives for the better. Usually this causes us to focus on more and better measurement. But maybe it is time to start thinking about better connecting our daily workflow to a computer that can organize all this information and share findings with us in the moment.

Leave a Reply