At EMC World in May, we presented EMC IIG’s solutions for optimal decision making (see my earlier post). The anatomy of optimal decision making we state requires these essential component: context, perspective and insight. In this post I talk a little more about the need for context in case management.
I always feel a little weird talking about context because if I get too deep, I fear I will sound too academic, theoretical or ostentatious - I don’t want to do that. The fact is, people consider context in the process of decision making all the time. Actually, the process of gaining context starts earlier than that - it starts when you truly want to understand communication. How much context you need and the process of gaining it vary. Context is required to make good decisions in simple and complex scenarios - from what to wear to a customer meeting, to big-time purchases, from choosing a restaurant for a first date to deciding which job to take (if you are lucky enough to have options).
Often times, the lack of context leads to poor decisions and outcomes. Some outcomes can be easily fixed - such as returning an item or never going back to a restaurant again. Some decisions lead to more serious circumstances or grave outcomes that cause damaged reputation, financial loss, irreparable damage to a relationship or loss of life. I think it's the latter that keeps most folks up at night.
Remember, even with the "best" information and all the context in the world, we can still make poor decisions. After all, we are human and machines are...well not human.
Context, what is it? And why do I need it?
One definition states:
"the set of circumstances or facts that surround a particular event, situation, etc." - dictionary.reference.com
I'm going to modify this a bit to say:
"the set of observations or relationships that surround a particular thing"
I think we can all agree on this, yes?
Let’s look at an example in case management. In investigative cases, observations and relationships are what drive the investigation. When a crime is committed, investigators need to know the physical characteristics of the alleged perpetrator - which prepares them to visually (or audibly) to identify the person of interest. And they also need to know the relationships the perpetrator has with other people, organizations, places, etc. They'll often ask a witness to "describe the assailant" or "who do they know" or "where do they work". The answers - the observations and relationships - are the context. With this context, investigators can then make the best decision possible on what to do next in the investigation. The context is intelligently driving the investigation - the processing of the case.
Note my choice of the word "observation". I want to be clear that these are not necessarily facts - and often they aren’t. A witness may have observed the subject of interest to be a certain height, weight, sex or ethnicity or that they saw the perpetrator with someone and inferred they know each other; but in a lot of cases observations do not turn out to be facts. Granted, we often take observations as the truth.
The above scenario probably reminds you of the popular TV Series, Law and Order. Where the inspectors are gathering and disseminating interrogations, witness accounts and other information on the fly and with great accuracy. But the reality is, in most cases the context and relationships are "hidden" in unstructured content - in documents as well as images, audio and video. Context is waiting to be discovered or extracted. In a given case, numerous pieces of content are collected that contain the observations (or facts) of a subject of interest.
The “why” should be obvious, right? We need to make optimal decisions, and we can’t make them without context.
Well, well...In just a few paragraphs, I've insulted machines (well, maybe they don’t really care), I've changed the definition of context to make my case, and I have even referenced a TV show. Time to “real” it in.
I like context. How do I get me some?
If you expand your perception of case management beyond the legal or criminal, and include any case that requires "digging deeper", you'll encounter in most instances a large volume of unstructured content and most likely unstructured processes - since context is needed first to help drive the processes.
This content holds a wealth of context that you don’t readily realize unless you analyze it using some software or ask human information workers to read all of it and tag it. You can certainly try the latter, but it becomes unscalable rather quickly (not to mention the issues in ensuring consistency)
So, how do we solve this issue of getting context out of massive amounts of content?
Well, you need content analytics - and as your content continues to grow (and these days the growth is astronomical), the importance of having content analytics grows as well.
At a minimum you need to be able to:
- discover entities (using a NLP-based approach) in content like people, places, companies - when searching later, you can find cases that share entities and with an integrated knowledge base, you can get context on the entities themselves
- categorize content and cases with a business taxonomy - entities in the content don’t necessarily tell you what the content is about, categorization does
- discover metadata by rules - sometimes you just know what you are looking - if you are familiar with the case content and your domain is tightly-scoped, rules help you get your context faster and with precision
- discover relationships - or rather identify subject-predicate-object triples in the content and creating the opportunity for human knowledge workers disseminate and manage relationships within and across cases - leading to richer context
Now, instead of just having a massive amount of strings of characters, you have context. You can now derive meaning from them at a scale much more suitable in an environment where intelligent decisions need to be made in an efficient and compliant manner.
What happens when you don't have Context?
Making decisions without context puts the decision maker at a severe disadvantage. Maritz Research published a research whitepaper: Context Before Cool Charts in 2010 that identifies two mistakes researches make when they don't have context. (I copy part of it here; but follow the link above for the entire paper):
- They mistake the norm for insight
- They miss signals and flags
If you've gone through my presentation - Turning Information into Business Advantage , you'll remember: 80% of business leaders make major decisions with incomplete or mistrusted information (IBM's Insitutute for Business Value 2009). You can confidently point the finger at lack of context (and perspective; but I'll get to that in another post) as the culprit. Without context, what people have is a set of obervations or facts - with seemingly no relationships to anything else - that don't mean much - and without anything else to go by, this becomes insight.
Scary.
“Wow” rather than “cool”.
Content analytics is great. I’ve been involved with designing and implementing solutions leveraging content analytics for a number of years. Initially, we were enamored by the recall - and the “cool” things we found in the content. But, you have to look beyond the “cool factor” and figure out the business advantage of content analytics in a real-life solution or application that provides real business value - and I think we’ve done that at EMC by integrating content analytics with content and case management. Content analytics is not the end - the end is the optimal decision. In case management this is especially true as we want to avoid making poor, uninformed decisions that cause grave circumstances.
Thanks for reading,
Batla