We all know how difficult it is to make a clear distinction between data, information and knowledge. Clear absolute frontiers between them can t be designed, at least in a common consensus. Literature gives almost as many different definitions as there are documents. I want to illustrate my vision on these topics by a rather developed example. I will take for this a very simple subject: "the Whiteboard Marker in a Discussion Room". Various scenarios will be given. After that we will have a look where the knowledge, information and data are and try to give a YaD (Yet another Description). Or perhaps this isn t a new one. Do I know all the definitions given over time?
Imagine the following situation. A particular room in a building serves for workgroups as a discussion room. This room is equipped with a whiteboard. One of the office employees is in charge for checking the room before a meeting. One of the tasks is to check whether there are whiteboard markers in different colors. For this example we presume that the markers that are there are still operational (not empty).
Scenario 1
The employee checks (at the end of the day) the day before the meeting the markers. If there aren t any he/she will go to the furniture service to get some and will bring them to the room. A few minutes before the session the meeting coordinator will come into the room. He perceives in a glance the state of the room and will go on with his activities. There seems nothing wrong in this scenario.
Scenario 2
The employee didn t execute the task, or the markers have "disappeared" after his check. In any case when the responsible person enters the room just a few minutes before the meeting, he takes the same quick look at the room and remarks that there are no markers. He will take action in order to get some (he might get them from next room or find somebody to get new ones from the furniture service).
A bit later in time this same discussion room is equipped with smart cameras in a way that all the cameras together "see" every corner in the room. They can communicate with each other and they have access to the corporate knowledge base. The employee formerly in charge of checking the room does now some other more interesting tasks and a software supervisor agent is in charge of the checking activity.
Scenario 3
Triggered by the corporate scheduler the new supervisor starts the check of the room. He knows that he must check the room on the presence of whiteboard markers. He asks one of the cameras "Do you see any whiteboard markers?" As the cameras are brand new, they don t know what markers are. The camera does a search in the KB and discovers that whiteboard markers are "things" bought from a particular provider (with an article reference). He will contact the provider and asks him whether he can give a 3D description of that particular article. We presume the provider is a modern company and will return the data asked for. The camera will now know what to look for. The camera knows that markers can be at several places in the room. They can be on the marker support of the whiteboard itself, on the tables, on the chairs and even on the floor. Together with other cameras the room is scanned and markers are discovered or not. The result will be communicated to the supervisor. If there aren t any the supervisor will ask one of the multipurpose robots to get some and bring them to the room.
Scenario 4
The start of this scenario is the same as the previous one but the camera can t discover anything about markers in the KB. So he don t know what he is supposed to be looking for. He will scan together with the other cameras all the objects in the room and tries to find out what they can discover about them. Whiteboard markers have most of the time text on them indicating that they are that particular kind of markers. Eventually with the help of one of the multipurpose robots that might reorient the marker so the cameras can read what is written on it and recognize it. If there aren t any, the cameras will decide that they can t say if there are any or not because they don t know yet what they are looking for. In case they can t say whether there are any markers the supervisor will ask a human being to give them some help to get this task executed.
Not all the aspects of this example will be detailed. You can work out some of them yourself, other s are of little interest in the actual explanation. In the situation 1 (scenarios 1 and 2) the employee knows how to recognize a whiteboard marker or if he doesn t yet he has the intelligence to discover how they look like. He also knows what his task is and how to execute it. Or perhaps he has this information on a checklist for preparing a discussion room. We might say that when he started to have the responsibility for checking the room he had no knowledge about how to do the check. Next either somebody told him, or gave him the checklist. Or he "discovered" by himself what should be involved in the task. We might say, and this only opens the discussion, that the checklist is information, the reading of the checklist adds some knowledge to the employee s brain. This knowledge, together with other knowledge, can be applied to execute the task. Next lets have a look at the execution phase of the task. When the employee checks the room his eyes will perceive an image (data?). After processing this data he will know that there are (or aren t) any markers. He will act according to this. The responsible of the meeting, when entering the room and having a quick look at it, will notice that there are markers. But he will probably not take consciously the decision that he doesn t have to do something particular with it. We might say that he only got the data (image) and processed it unconsciously. On the other hand if there are no markers, this information (knowledge?) will become important enough to get to his active mind so that he can take a decision for action. So the same data/information/knowledge (whether there are or aren t any markers), depending on the context (employee in his checking activity or the responsible person in his pre-executing activity), remain at the stage of data, grow to the level of (useful) information or even become knowledge (that implies an action). In this perspective data might be something like bits and bytes with no particular value added to it (the image the eyes receive). Information might then be data to which there is some value added (there are markers). And knowledge? Have a look at the descriptive text of the scenarios. When did we use the term "know"? Let s have a closer look at the second situation (1-1 with the scenarios 3 and 4). When we declare a fact by the means of the term "knows" we say in fact that we have provided these agents (supervisor or camera) some information that they can use to handle. They know also how to handle if they don t know something. When persons were involved (situation 1) the term "know" wasn t necessary. We took, quite naturally, for granted that people know things. In situation 1-1 it seemed to be necessary to indicate that the actors in these scenarios had to know some things in order to be able to decide what to do. So might knowledge be the whole of information that, linked together, allow a decision to be taken or to act?
The descriptions (I don t like to call them definitions) of data, information and knowledge given above don t have a large geographical, cultural or temporal validity space. But let s accept them for the time of reading the rest of this paper. The reading and writing mechanisms of Notion System will only be handling data (the bits and bytes). The core will manage information (Paris is located in France e.g.). But where is the knowledge? The semantic network (all the pieces of information linked together) provides human users or agents enough information to take decisions or to act. So I do think that the term "knowledge" is appropriate in the context of a semantic network. Whether the information from Notion System will really be used to decide or act (and thus upgrading the information to the level of knowledge) is beyond the scope of the system itself. Users (human or agents) of the semantic network might take this step.