Entry
This article is intended for those who are familiar with such a concept as ontology, at least at the initial level. If you are not familiar with ontologies, then most likely you will not understand the purpose of ontologies and this article in particular. I advise you to familiarize yourself with this phenomenon before you start reading this article (perhaps even an article from Wikipedia will suffice).
So Ontology is a detailed description of a certain subject area under consideration. Such a characterization should be given in some clearly articulated language. To describe ontologies, you can use the IDEF5 methodology, which has 2 languages ββin its arsenal:
- Schematic language IDEF5. This language is visual and uses graphic elements.
- Text language IDEF5. This language is represented as structured text.
This article will consider the first option - a schematic language. We will talk about text in the following articles.
objects
In a schematic language, as already mentioned, graphic elements are used. To begin with, we should consider the main elements of this language.
Often, ontology uses both generalized entities and specific objects. Generic entities are called species. They are depicted as a circle with a label (object name) inside:
Views are a collection of individual instances of a given view. That is, such a view as "Cars" can represent a whole set of individual cars.
As the copies this type can be specific cars, or certain types of equipment, or certain brands. It all depends on the context, the subject area and its level of detail. For example, for a car repair shop, specific cars as physical entities will be important. To maintain some statistics on sales in a car dealership, specific models, etc. will be important.
Separate instances of views are designated similarly to the views themselves, only they are indicated by a dot in the lower part of the circle:
Also, as part of the discussion of objects, it is worth mentioning such objects as processes.
If views and instances are so-called static objects (not changing over time), then processes are dynamic objects. This means that these objects exist in a certain strictly defined period of time.
For example, you can select such an object as the process of manufacturing a car (since we are talking about them). It is intuitively clear that this object exists only during the actual production of this very car (a strictly defined period of time). It should be borne in mind that this definition is conditional, because objects such as a car also have their own service life, shelf life, existence, etc. However, we will not go into philosophy, and within the framework of most subject areas, it can be accepted that instances, and even more so species, exist forever.
Processes are displayed as a rectangle with a label (name) of the process:
Processes are used in schemes for the transition of one object to another. More on this will be discussed later.
In addition to processes, such schemes use logical operators. Everything is simple enough for those who are familiar with predicates, Boolean algebra or programming. IDEF5 uses three basic logical operators:
- logical AND (AND);
- logical OR (OR);
- exclusive OR (XOR).
The IDEF5 standard (http://idef.ru/documents/Idef5.pdf - most of the information from this source) defines the representation of logical operators as small circles (compared to views and instances) with a label in the form of symbols. However, in the development of the IDEF5 graphical environment, we have departed from this rule for many reasons. One of them is the difficult identification of these operators. Therefore, we use the text notation of operators with an identification number:
Perhaps this is the end of the objects.
Relations
There are relations between objects, which in ontology mean rules that determine the interaction between objects and from which new conclusions are obtained.
Typically, relationships are defined by the type of schema used in the ontology. scheme is a set of ontology objects and relations between them. There are the following main types of schemes:
- composition schemes.
- Classification schemes.
- Transition schemes.
- Functional diagrams.
- Combined schemes.
Also, sometimes there is such a type of schemes as existential. An existential schema is a collection of objects without relationships. Such diagrams simply show that a certain set of objects exists in a certain subject area.
Well, now in order about each of the types of schemes.
Composition schemes
This type of diagram is used to represent the composition of an object, system, structure, etc. A typical example is car parts. In the most enlarged composition, the car consists of a body and a transmission. In turn, the body is divided into a frame, doors and other parts. This decomposition can be continued further - it all depends on the required level of detail in this particular problem. An example of such a scheme:
Composition relationships are displayed as an arrow with a tip at the end (unlike, for example, a classification relationship, where the tip is at the beginning of the arrow, more on that later). Such relationships can be signed with a label as in the figure (part).
Classification schemes
Classification schemes are intended to express the definition of species, their subspecies, and species instances. For example, cars can be cars and trucks. That is, the "Car" view has two subspecies. VAZ-2110 is a specific instance of the subspecies "Car", and GAZ-3307 is an instance of the subspecies "Truck":
Relationships in classification schemes (subspecies or specific instance) have the form of an arrow with a tip at the beginning and, as in the case of composition schemes, can have a label with the name of the relationship.
Transition schemes
Schemes of this type are necessary to display the processes of transition of objects from one state to another under the influence of a certain process. For example, after the process of painting with red paint, a black car becomes red:
The transition ratio is indicated by an arrow with a tip at the end and a circle in the center. As you can see from the diagram, processes refer to relationships, not objects.
In addition to the ordinary transition shown in the figure, there is a strict transition. It is used in cases where the transition in a given situation is not obvious, but it is important for us to emphasize it. For example, mounting a rear-view mirror on a car is not a significant operation if we consider the process of assembling a car globally. However, in some cases it is necessary to select this operation:
A strict transition is denoted similarly to a normal transition, except for the double tip at the end.
Regular and strict transitions can also be marked as instant. To do this, a triangle is added to the central circle. Instantaneous transitions are used in cases where the transition time is so short that it is completely insignificant within the subject area under consideration (less than the minimum significant time interval).
For example, with even the slightest damage to a car, it can be considered damaged and its price drops sharply. However, most damage occurs instantly, unlike aging and wear:
The example shows a strict transition, but you can also use a normal transition as an instant transition.
Functional diagrams
Such schemes are used to indicate the structure of interaction between objects. For example, a car mechanic performs car maintenance, and a car service manager takes requests for repairs and passes them to a car mechanic:
Functional relationships are depicted as a straight line without a tip, but sometimes with a label, which is the name of the relationship.
Combined schemes
Combined schemes are a combination of previously considered schemes. Most schemas in the IDEF5 methodology are combined, since ontologies that use only one kind of schema are rare.
All circuits often use logical operators. By using them, you can implement relationships between three, four or more objects. A logical operator may express some general entity on which a process is performed or which participates in another relationship. For example, you can combine the previous examples into one as follows:
In a specific case, the combined scheme uses a composition scheme (mirror + car without a mirror = car with a mirror) and a transition scheme (a car with a mirror becomes a red car under the influence of the red paint process). Moreover, a car with a mirror is not expressed explicitly - instead, the logical operator AND is indicated.
Conclusion
In this article, I tried to describe the main objects and relationships in the IDEF5 methodology. As an example, I used the subject area related to cars, since it turned out to be much easier to build diagrams on their example. However, IDEF5 schemas can be used in any other area of ββexpertise.
Ontologies and domain knowledge analysis is a rather extensive and time-consuming topic. However, within the framework of IDEF5, everything turns out to be not so difficult, at least the basics of this topic are learned quite simply. The purpose of my article is to attract a new audience to the problem of knowledge analysis, albeit at the expense of such a primitive IDEF5 tool as a graphic language.
The problem of a graphic language is that it cannot be used to formulate certain relations (axioms) of the ontology clearly enough. To do this, there is a text language IDEF5. However, at the initial stage, a graphical language can be very useful for formulating the initial ontology requirements and determining the vector for developing a more detailed ontology in the IDEF5 text language or in any other tool.
I hope this article will be useful for beginners in this field, maybe even for those who have been dealing with the issue of ontological analysis for a long time. All the main material of this article has been translated and understood from the IDEF5 standard, which I referred to earlier (). I was also inspired by a wonderful book from authors from KNOW INTUIT ().
Source: habr.com