Knowledge and … toothaches
In most cases, the use of language, written or spoken, has one goal: to convey information. Our world is crossed by a myriad of news streams in constant interchange, and in intersections along a network that envelops the entire planet.
Contrary to what happens in the perception of sounds, colors or smells, we can’t sense or experience knowledge in the same way. Instead, the brain, which more so than the eyes, ears and nose, allows the mind to “know” the outside world and understand its meaning.
That is why the meaning of a text cannot be inferred mechanically by understanding the individual words, but is always the result of an interpretation of the context in which these words are used.
“Understanding”, “knowing”, “communicating” and “learning” are all human actions, even if they seem innate or natural; we “know”—full stop. It is more difficult to explain how we know, but even so, we succeed in knowing because we have correctly interpreted actions and concepts, information.
To be successful in the field of information means capturing the meaning and locating just the right information, discarding what is less relevant (and therefore not wasting time or resources). Our daily challenge (in any business), is being able to use only the knowledge we need, when and where we need it, to derive maximum value from it (which often means making decisions). In reality, this is a complex process requiring successive refinement as we climb the ladder of increasing levels of information.
Consider a simple example of searching for information. In general terms, the most basic level is data. Suppose we select a simple sequence of 10-digit numbers. Our first sign of comprehension transforms them into weak information and we are able to identify them in context: they are phone numbers.
Climbing to the next level increases the content of our information: now that we know they are phone numbers, we can also recognize that the first three digits as area codes, which we assign, through our contextual knowledge, to a certain region. We reach the maximum information value when the context becomes so close to our goals that we can base a decision on it: We recognize some of the prefixes that correspond to a university, a hospital or a community. If I need to find a dentist in my city, for example, this could narrow the numbers down to what is relevant for me.
The same path—from data to knowledge—is also made in the automatic processing of information (in our case, text) performed by a program. In this case, a technology can be considered valid if it is not limited to operating in only the early stages of the scale by recognizing and manipulating data or general information. With almost no understanding of the context, and rough surface, a technology must be able to go further and, as the human brain does, distill the information down to what is relevant knowledge.
Even if the level of human understanding has not been replicated (to date), semantic technology provides a close approximation through its ability to understand the meaning of a text and analyze the concepts and their interconnections.
What a text expresses is not just the simple sum of the meanings of the individual words. Instead, the meaning is clear from the concatenation of words in sentences, by the succession of phrases in sentences, and often from our pre-existing knowledge.
All of these elements are interwoven like threads in a tapestry, which make up the overall design. It’s no coincidence that the word “text” derives from the Latin word “tessuto” or “fabric”.
For now, I will stop here. In the next post, we will see how a computer, born “only” to do calculations, can do so much more.