Navy Real-time Knowledge Sharing (RKS)

MARi, LLC and Carnegie Mellon University are partnering to combine state of the art multimedia content tagging algorithms together with granular levels of task and Sailor analytics to ensure that knowledge objects can quickly move through a semi-automated, custom approval pathway - getting the latest and best information to the right person at the right time.  Knowledge sharing systems designed for performance improvement in the workplace have largely proven ineffective due to the significant amount of labor required to maintain high levels of “real-time relevancy” of the information for the end user.  Without a significant amount of work to keep information relevant, the eventual problems of too much information, not the right information, and information that is not validated as correct – creates a cycle that stops the use of the system from both the people that input the knowledge objects (KOs) and those who need or are seeking the knowledge.  In short, it becomes a waste of time. In the workplace, however, where increased productivity and accelerated performance are the required outcomes, the standard for the relevancy of the shared KO is extremely high.  In a Navy peer-to-peer knowledge sharing system, extremely high relevancy must be equally paired with high trust in the information – both for the end user and the entire chain of command.  The combination of high relevancy and trust has traditionally been achieved through a sometimes-lengthy curation process that approves any KO for distribution.  The current and future Navy requires a much faster way to get real-time, vetted, reliable knowledge from the deckplate to the rest of the fleet as quickly as possible. The MARi Team will be applying an advanced “translational” model vs a “standards” model to the knowledge objects to both move tagged KOs through a permissions-based gating process and establish high relevancy of the knowledge objects to the end users.  Traditional systems, even those that use AI and machine learning (ML), use a tagging system that relies on some type of common library of content terminology or a set of standards for how to describe something.  In almost all cases these standards contribute to the labor overhead that causes knowledge sharing systems to collapse.  In a translational model, our system will continuously be learning which input tags translate into high relevancy tags.  When our translational model is combined with granular data about the Sailor that is entering in the KO and the Sailor that is consuming the KO, we can achieve the high-trust and high-relevancy required to ensure the knowledge sharing system is sustainable.