(Unstructured) Big Data: Context is King
Structured and unstructured content require a separate approach to manage and extract business value. The large amounts of structured data requires a high performance computing solution. The unstructured data requires a solution with a similar ability to scale, but for the solution to be effective, it must possess a unique ability to effectively identify context. Structured solutions do not require word disambiguation in the same way as unstructured solutions.
People are able to disambiguate on the fly, (ex: you are able to quickly understand what I mean by “on the fly”), but machines cannot. Stop for a moment and do a web search. Check the results. Key word and statistic based approaches in isolation are not as effective as they could be. If context had been established during the search, the results would have been better. They would have been more precise, they would have been more complete.
People disambiguate by evaluating context, machines cannot. People disambiguate through the use of learned understanding of words and or word combinations, machines cannot. If I say the word “stock” there is no context. For a person to understand context, they look at the words around the word (context clues) or ask for more information to establish context. Even when a word meaning is unknown, context can be established and meaning can at times be determined by the words around the word in question. This is not a machine capable operation. If I combine words is the following way, “I bought 10,000 shares of stock in Apple.” context has been established and meaning is understood. If I combine the same main words in a different order, “I have 10,000 apples in stock.” the word relationships change, therefore the meaning changes. This is very important and a critical piece to the understanding of language. When this unique ability to understand context is successfully transferred to a machine we have context and this unique ability of establishing context drives value.
Scalable technology is critical when working with big data, but context is king. For computers to deliver context, it requires a knowledge management platform that delivers word disambiguation through language understanding.