Declara Cognitive Graph
You’ll hear this word in our vocabulary called The “CognativeGraph”. You’re probably asking what is a CognativeGraph, what does it even mean? To describe it in simple terms, the CognitiveGraph is a very expandable, scalable set of algorithms. The CognitiveGraph is machine learning, data algorithm, which is the secret sauce, IP, of Declara.
If you think about search in the sense on how we go about looking for things, you’ll know that some people are better at searching for content, than others. The Declara CognitiveGraph aids people in their search for content, by monitoring the five inputs. Intent, Interest, Social, Knowledge & Reputation.
Intent is a short term need to learn something
This is where we start to tear apart what the true intent was. Intent and Interest sometimes overlap and sometimes not, as interests are things I like. For example, sometimes you have to do something for work (intent), but you’re not really interested about it. Intent is behavioural knowledge over a short period of time, while interest, true interest, is built over a longer period of time.
This takes into account your social graph. Who are the people that you interact with, and who are the people interacting with you. What is the ratio of that interaction, what are the things you’re interacting on with those people. How successful was the content. Was it shared, did you like it, did others like and share it.
Knowledge simply acknowledges how much do I know about this subject. You can watch 1,000 cat videos but that doesn’t make you a vet. As you build your profile and content in Declara you breadth and depth of knowledge in your intent and interest grows.
It is not enough to know, it is also very important to have influence, that people are interested in what you have to highlight. You may not be the original curator of that content, but you’re a very important filter that brings up curious intent which is high quality.
The CognitiveGraph is what Declara relies on to deliver content, the knowledge that people need at the relevant time. To help people know what they don’t know
From the 5 inputs Intent, Interest, Social, Knowledge & Reputation, we can come to two points of intelligence; First type of intelligence, is predictive intelligence. So how do we showcase predictive intelligence to the end user, we do it by our recommendations feed. This is what powers our recommendation engine. That’s over simplifying it, but it’s very complicated. By using the predictive intelligence we create high quality recommendations.
The second part is transformative intelligence, and we surface this to the end user by a Learning Profile. What the learning profile contains
#1 Depth of Engagement and
#2 Breadth of influence, we catch these two in the learning profile.
Your learning profile is like your social profile on Facebook, or your professional profile on LinkedIn, it’s evolving everyday, it is the reflection of your learning process, and it’s completely portable, as you go from one job to the other it follows you. The collections you build in Declara is like your “naked brain”
The CognitiveGraph algorithm also pays attention to:
Did you read this on a computer, did you read this on a tablet etc. Are you someone that prefers images, or the insight, did you follow the person etc.
All these behaviours start to initiate to build a dynamic archetype of how you like to learn, and how you are as a person.
These could be things like, how fast is the mobile device moving right now. So if it’s moving fast, instead of sending you a text link, then we send you a audio message.
Successful / Failure Gate
One of our early learnings was that there were some consequences with unintended learnings. For example if I wanted to learn more about historical wars, I’m bound to have articles on the type of artillery used, but that doesn’t mean I want guns in my feed. Therefore guns would be noted as a failure for content, and won’t appear in your feed.
To ensure you find exactly what you were looking for Declara has:
• Search — We have a search functionality. It’s very elastic, and we’ll search not only by text, but relevancy to your interactions and behaviours you have.
• Insights — We want to be able to send you the insights that you find interesting.
• Curation — The ability for you to curate something, your collection on a subject that your researching
• Collaboration — Your ability to network all of your collections socially, so there’s the collaboration enabelling group or team based learning