How to confuse analytics. Part three. Verbs and Numerals

    In a previous article, I proposed an approach to modeling nouns and adjectives in such a way as to obtain a repository of a subject area that does not require a change in its structure when new knowledge is added to it. It turned out the following:

    • We model accounting objects
    • Using nouns and adjectives, we model their classification.
    • Classification is always subjective and therefore it is necessary to indicate the subject who made this classification

    I conducted thought experiments on combining the repositories in order to check the created structure of the repository for its spaciousness: I studied what can be included in it and what can no longer fit without altering the structure.

    We sorted out nouns and adjectives. I would like to deal with verbs just as easily. However, with verbs this simply does not work out.

    First, we clarify our ideas. Until now, we considered our world to be static. If it is a car, then forever, if it is yellow, then forever. In reality, of course, this is not so. The same accounting object at different times by the same entity can be classified in different ways. This is the consideration of the life cycle of an object. Therefore, the objects of accounting that we model in IP are not objects in general, but their temporal parts. For example, we model not the entire car that existed from 1990 to 2010, but only its temporal part - the state of the car when it was yellow: from 1995 to 1999. The temporal parts of the accounting object are also accounting objects. This means that any accounting object can be divided into parts, which are also accounting objects.

    What is behind verbs?


    Saying: “Martynov sold the machine” it seems obvious and natural. Who! Martynov. What did you do? I sold it. What did Martynov sell? Machine. Just like in the 6th grade of the school. And it seems to everyone that there is nothing strange in this. It seems that the subject has performed the action, and this is normal. But the question is: can the subject perform actions? And this is not a philosophical question, to which Descartes answered. This is a question from the field of modeling. Can I build models based on the hypothesis that objects and subjects perform actions, for example, can I assume that the car is traveling?

    I asked a similar question in a previous article: can I build models based on the hypothesis that objects have attributes? The answer was no. This is explained simply: if an object has a property, then you might think that the object has a property by nature. However, the subject gives it its properties. At the same time, different entities will endow one accounting object with different properties.

    In order to understand whether it is possible to endow objects with properties, I conducted a thought experiment to combine two information repositories. Verification by combining showed that two repositories can be merged when attributes are separated from types and their values ​​from objects.

    If we talk about verbs, then the analogy is the same: only the subject can say who was the performer of a particular action, while different subjects can give different answers to this question. Therefore, there is a hypothesis that the performer cannot do anything in the same way as an object cannot possess a property.

    Let's continue the thought experiment.

    Suppose there are two repositories of models, in one of which is written: Martynov sold the machine. The second store says: Gavrilov bought the machine. How to combine storage so that it is clear that it was one operation?

    In OOP, actions are modeled by methods. A class is created whose objects are capable of doing something. For example, a class of people. Objects of this class are able to sell and buy. And there are two records that Martynov performed the sale method, and Gavrilov completed the purchase method. To combine these two records into one, there must be one record that looks from the point of view of Martynov as the execution of the method to sell, and from the point of view of Gavrilov as the execution of the method to buy. To do this, we can create a method whose participants we will make Martynov, Gavrilov and the machine, classify this method for Martynov as a sale, and for Gavrilov - as a purchase. Thus, we will equalize the degree of participation of participants in the operation by denying the participants the right to do something, at the same time invented functional programming.

    It is very important: if we want to build extensible repositories of models, we will have to take for granted that there are no executors of operations, but there are participants in operations. The contractor is nothing more than one of the many interpretations of the participation of one of the participants in the operation.

    Once a lecturer was late for a lecture. It was in Japan, where being late is a serious violation of etiquette. I was wondering how he would get out, but he only said: “I did everything I could. The circumstances were higher than me. ” So in any operation - we can think of anyone as a performer, but in reality it turns out that nature is stronger.

    A little more carefully examine the concept of operation. An operation is an accounting item. Only unlike “static” metering objects such as a machine, is the operation dynamic and fleeting. On the other hand, “static” objects, for example, Martynov and the machine, can participate in the operation. In other words, there is an accounting object, the parts of which are Martynov and the machine, which is interpreted by the author of the model as a sale operation. But part of the operation Martynov and the machine were not completely, but only partially. The operation lasted only one hour, and all participants live much longer, so Martynov and the machine tool, and their temporal parts, are not involved in the operation. Therefore, the operation model looks like this: there is an accounting object, interpreted by the author as a sale operation. This accounting object includes the temporal parts of accounting objects, which are interpreted as Martynov and the machine tool. The temporal part of Martynov, which is part of the accounting object treated as an operation, is treated as the performer of the operation. All interpretations are subjective. Such an operation model is extensible. If her new interpretations appear, for example, as a purchase, or new participants, for example, Gavrilov, it will not be difficult to add them to the model.

    We got the following: for modeling verbs, it is necessary to create a model that takes into account the relationship between objects of accounting of the "part-whole" type.

    Numeric attributes


    It is often thought that lengths, time intervals, weights and other properties expressed in numerical values ​​are natural properties of objects. But in fact, there are no properties outside our consciousness. All these attributes, units and methods of scaling exist in our minds and depend on the tasks we solve.

    For example, if you are an artist, then the color will be conveniently modeled using names. If you are a physicist, then, depending on the field of physics, you will model the color using a comparison with the standard:

    • Lengths (wavelengths if you are doing wave optics)
    • Frequencies (wave frequencies if you are a radio physicist)
    • Energies (quantum energies, if you are involved in quantum optics)

    If we are talking about scales and numbers, then in addition to the standard, we must choose a method of comparison with this standard. For example, you can ask a question: how many times is that tree above the reference meter? It seems like ten times. But this is the apparent evidence. In fact, you could use a different method for comparison. For example, use the logarithmic scale. Then the tree would be twice as high as a meter.

    The choice of a comparison method with a reference depends on the problem being solved. At the same time, operations on attribute values ​​will simulate operations that we perform on real-world objects. For example, the addition of the masses of two bodies makes sense when it is necessary to find out the mass of a body composed of two bodies. However, the addition of wavelengths of two light quanta is an operation without meaning. But you can add the energy of these quanta to get the amount of radiated energy.

    Without an answer to the question about the standard, the method of comparison with it and without a description of what operations on objects are modeled by what operations on the numerical values ​​of attributes, the attribute value itself is meaningless. It’s correct to say this: the numerical value of the attribute is obtained by comparing it with a standard (unit of measure) performed in the indicated way as part of solving certain problems.

    And then you will know, for example, that the height of the building is measured from the zero mark, which coincides with the level of overlap of the first floor. You may learn that to describe the properties of an electron called spin, numerical values ​​are no longer enough, and you have to turn to matrices. And, perhaps, you will find out that in addition to one-dimensional scales, there are multidimensional and even infinite-dimensional scales.

    Also popular now: