We’ve introduced the concepts of the data types already. From these we will build a better front end system for capture to produce the required data to solve our problem.
· RD: Raw data, available but may not include schema, context or connection
· PD: processed data, some connections, schema and context
· ID: Indexed data or data with context assigned
· AD: Available data, which is a good system would be the other three combined.
Given these four data types and incorporating the concepts of John Boyd’s OODA Loops, including the broader feedback loops we will position the data we are considering into our framework and build a schema for inputting data.
First off, one of the huge problems for KM systems has always been ingest. Ingest is the process required to get data from the raw to the indexed effectively. With our four types of data being considered we can build a quick ingest formula.
Discover builds on the concept of all data regardless of origin. Evaluate allows us to apply criteria including the various feedback loops we will build and assimilate allows us to move data into a more rapidly reusable location.
Within each of the three there are also grades of application, for example if you are making a decision right now, having a broad 2-3 days’ discovery sweep of available information isn’t going to help you. So we balance the three against the time requirement. Our initial feedback loop then is how quickly do we need to get and use the information. This is why we have an all-data tag. Sometimes you need to quickly search all data to find what you are needing.
Data Feedback Loop 1: How fast do I need the data.
Our next feedback loop to consider is the value of the source. Trusted sources give us known good data, or data we trust. It doesn’t mean the source is always right, just that the information the source provides is trusted. Our second feedback loop then is the application of an initial source rule. For example, asking a doctor about the workings of a cyclotron may not provide the results you are expecting. The doctor may know a little, but its highly unlike unless he or she is a doctor of physics, that in fact they know enough. Medical doctors for example wouldn’t have the information required. So we evaluate sources only on past performance.
Data Feedback Loop 2: Compare the source to the value of the information provided
Finally, there is the concept of assimilation. Not all consumed data needs to be moved for formal storage systems for later reuse. Not all data should be released into the wild however once consumed. Our loop here is to find and focus on the rate of occurrence of the problem. If the problem is a one off, we consume the link to the data but not the data. If the problem is occasional we consume a synopsis of the information and a link to the original but not the entire thing. If the problem is chronic, reoccurring and a huge issue, we consume the entire thing (with permission of course).
Data Feedback Loop 3: Apply time requirements to data capture.