Universal Transcendental Logic-Based Ontology

In this paper, Kant's philosophical doctrine of the categories of the reason is used to substantiate the conceptual model of knowledge representation, based on the collective interaction of a lot of intellectual atomic elements of knowledge (knowledge quanta), which are combined into clusters like neurons in the brain; and also a phenomenological description of the corresponding universal ontology, proceeding from the philosophical premise of Husserl-Heidegger that the meaning of intelligence is not so much in knowing the absolute truth as in survival, is presented. In the process of cognizing the surrounding world, a person uses both a priori knowledge and a posteriori knowledge, but the transcendental content of a priori forms of thinking does not allow them to be used directly in logical judgments. Nevertheless, one can try to use them as "ontological predicates" following the advice of I. Kant, what was done in this article. Heuristic ontological relations that directly follow from the categories of Kant are easy to use and sufficient to describe any ontology. Offered knowledge representation model, the key idea of which is the primacy of knowledge to logical inference and their emergent ability to self-organize, in conjunction with the transcendental logic-based ontology of empirical knowledge can be used to create a universal inference engine.


Introduction
The stumbling block for creating artificial intelligence is the question of the origin of knowledgecognition. The origins of modern epistemology were laid by I. Kant's categories of reason. According to I. Kant, both the forms of thinking and the forms of perception can be a priori (pure) and empirical. He defined the a priori categories of reason, on which depends how we see and understand the world, and which are a priori abstract forms of thinking. To apply them to reality, that is, to go over to a posteriori forms of thinking, they must be superimposed on a priori forms of perception, which are time and space. It should be borne in mind that one cannot limit the cognition of the surrounding world by the conscious phenomenological perception in space and in time, forgetting about the possibility of the subconscious (heuristic) cognition through communication with the Creator. It is the presence of such a connection that distinguishes the human being as a special type of organic life on Earth from animals; without this connection, his full intellectual development is impossible.
us what properties an object has, but indicates how we should, guided by it, reveal the properties and connections of objects of experience in general" (Kant, 1922). Thus, transcendental logic concerning general logic has a "regulatory application", which Kant himself called a heuristic.
The idea of Kant's transcendental logic arose from the inadequacy of the apparatus of formal logic for solving the problems of metaphysics and the theory of knowledge, but he deduced the categories of reason from the theory of judgments, that is, from general logic, therefore there are as many categories of reason as there are types of judgments in general logic. It is not surprising that modern artificial intelligence systems based on modeling formal-logical inferences are limited in their capabilities for modeling the cognitive process. As noted in (Moroz, 2020), their limitations are due to the following reasons: 1. Knowledge is not decisive for intelligence, although, as commonsense reasoning suggests, intelligence is, first, knowledge, no knowledge -no intelligence. "Stupid as a child"that is saying about a person who lacks knowledge. The focus is made on logical inference, which is built on formal logic with all its shortcomings.
2. Knowledge is separate from thinking and is not active. In the architecture of these systems, two main components are distinguished: the inference engine and the knowledge base. At the same time, regardless of the knowledge representation model (decision trees, semantic nets, predicate calculus, and so on), the knowledge base is very limited, so that with its growth, its inconsistency also grows, that leads to the need to use another component in the systemthe explanatory interface. This division is a direct consequence of the rather widespread misconception that the brain is a computer and thinking is a program. That is not right since knowledge is an active substance, the forms of activity of which are inextricably linked with the process of thinking and, thus, with consciousness.
All the above applies equally to other models of intelligence, such as neural, genetic, or social, which are trying to overcome the rationalistic legacy of previous generations of artificial intelligence (Luger, 2009). They produce inferences not based on logical rules but based on a collective decision generated by a sufficiently large number of interacting agents. All of them are artificial narrow intelligence and follow pre-programmed rules in their functioning, so their decisions seem to be reasonable, especially in cases of "deep learning" where the system can learn and adapt really. A prime example of this is the use of Artificial Neural Networks to simulate symbolic computation, presented in the paper (Lample & Charton, 2019), where a neural network was used that "remembers the context", that is, it considers the meaning of words depending on which words are near to them. Meanwhile, the fact that the meaning of any utterance depends on the context of its use was argued by L.
Wittgenstein (Wittgenstein & Anscombe, 1953), who revised the concept of meaning in natural and formal languages. Like M. Heidegger (Heidegger;Macquarrie, & Robinson, 1962), Wittgenstein has repeatedly criticized the rationalist traditions in modern philosophy and science in his studies.
Despite the rejection of logical conclusions like the main mechanism of thinking, the achievements of such models of artificial intelligence, although advertised quite widely and ambitiously, are very modest in terms of artificial general intelligence, and sometimes their intellectual abilities lead to curiosities. They lack flexibility and ability to learn (cognition), in a universal human broad sense, and their application is limited by the framework of private ontologies.
In this paper which is the continuation of the paper (Moroz, 2020), Kant's philosophical doctrine of the categories of the reason is used to substantiate the conceptual model of knowledge representation, based on the collective interaction of a lot of intellectual atomic elements of knowledge (knowledge quanta), which are combined into clusters like neurons in the brain, and also a phenomenological description of the corresponding universal ontology, proceeding from the philosophical premise of Husserl-Heidegger that the meaning of intelligence is not so much in knowing the absolute truth as in survival, in knowing its place in this world, is presented.

Method
Understanding the relationship between general and transcendental logic is very important since both are rules for gaining new knowledge, deriving some concepts from others. The difference between them lies in the attitude to the content of knowledge: general logic is abstracted from any content of the concepts with which it operates, and transcendental logic deals with a certain content -"transcendental content". Because of the limitations of general logic, there is a natural desire to apply transcendental logic to model the universal mechanism of logical inference in artificial intelligence systems. However, here we are faced with a problem essentially since transcendental logic as an integral part of the mind is not directly applicable to a posterior knowledge. The role of transcendental logic "consists in limiting the set of all possible conclusions of the given judgments. Not all conclusions deemed valid by formal logic are valid from the perspective of transcendental logic; and what is more important, the limitations to space, within which the search for logical inference is conducted, are not arbitrary, but have clear ontological bases relating to Kant's limitations on the application of categories» (Bryushinkin, 2011).
Moreover, transcendental logic cannot be formalized, since it is inherently heuristic, although some researchers make attempts to construct a certain logical formalization corresponding to the Kantian categories of reason, based on geometric logic (Achourioti & van Lambalgen, 2011) and input/output logic (Evans, Sergot, & Stephenson, 2020), that is it is about the construction of intuitionistic semantics of transcendental logic. An alternative way of developing symbolic computation based on transcendental logic can be the introduction of ontological relations between concepts, especially because I. Kant wrote in one of his letters concerning the categories that "their logical employment consists in their use as predicates of objects" and called them "ontological predicates" (Kant, 1914). In (Bryushinkin, 2011), these relations are based on P. Strawson's descriptive metaphysics. In this paper, we have defined twelve kinds of relations, which heuristically resulting from the original Kant's categories.
As well as thinking, which is categorical, knowledge being a product of thinking is also categorical, that is, it consists of the concepts referring to certain categories (types, classes) and having the relationships that do not contradict the a priori forms of cognition and thinking and allow to evaluate the meaning of these concepts depending on the context used. So, we consider the structural atomic elements of knowledge such as referring to different categories connected by the ontological relations. Table 1 presents these relations and the corresponding initial Kant's categories. The first six relations express the meaning of the philosophical concepts of quantity and quality, cause and effect, and modality. They do not require additional explanation and are used to describe phenomena much more often than others. The same can be said for the attitude "Relation of variability" which describes the ability to change the concept itself that is its state or properties, for example, "flexible concept", "blue sky", "hard stone", "bird flies", "car rides" and so on.
"Relation of tactility" allows setting a degree of semantic proximity between different concepts, for example, "winding coast" (like a curve), "head engineer" (like a head), "wonderful evening" (like any surprise phenomenon).
perceptions, neither pure nor empirical. "Relation of vitality" defining the role in the life of the individual corresponds to this super-category. This attitude is always associated with the answer to the question "What is the most important thing in life?" or "What is the meaning of life?". The answer is always subjective and original in each specific implementation.
"Relation of wisdom" sets influence on mental abilities in a broad sense like stupid-trick, logical-illogical, reasonable-stupid, and so on, for example, "smart horse", "illogical actions".
"Relation of chaos" corresponds to the randomness of the established relationship. For example, «red elephant» this is not realistic but maybe in imagination.
"Relation of effectiveness" sets the ability to generate control instructions, the desire to be first, to make others dependent on yourself and others, for example, "soldier goes to the attack" (command "attack"), "stand in line" (command "stand").
Consider the following example: "Heating water causes it to evaporate." In this case, steam is a consequence of water heating, that is, the concept of "steam" is related through the "Relation of causality" to the concepts of "heating" and "water", but not to these concepts themselves, but to the connection between them, which refers to "Relation of variability". So, "Relation of causality" characterizes the causal nature just of the relationship between the concepts of "heating" and "water".
Consider another example: "The car is driving on the road." Here the concepts of "car" and "drive" are linked by the "Relation of variability" (this state of the machine is "driving"), and the concept of "road" additionally characterizes, but not the concepts of "car" or "drives", but their relation, that is the concept of "road" is related to the relationship between the concepts of "car" and "drives" using the same "Relation of variability".
And one more example: "Twilight, fog, sad ...". Here are no relationships between the concepts at allthey are linked by meaning, by context.
The ontological relations presented in Table 1 are easy to use and sufficient to describe any ontology. In this case, one should keep in mind a few simple rules for using them: 1. The ontological relations can express both connections between two concepts and between concept and relation. In the first case, the relation characterizes both concepts associated with it, in the secondthe concept acts as an additional characteristic of the relationship associated with it.
2. One can use concepts between which there is no ontological connection, but they can be related by the context of use, they can be united by the context in meaning. The meaning of any statement (not necessarily verbal) is determined by the mind. When we say, "this is meaningless," it means that something goes beyond the reasonable, contradicts the transcendental content of our thinking. In any situation, we check our common sense, not realizing that this is just our mind, but it depends on the situation and on that who makes the decision.
3. Relations can be nested, and the depth of nesting is limited only by common sense.
4. And, finally, when applying these relations to construct an ontology, one should not forget that the starting point for them is not the establishment of ultimate truth, but the description of knowledge gained from experience, usually by trial and error. Figure 1 shows the ontological diagram for the chain of thinking corresponding to the utterance "The mountain was very high and overgrown with forest". Recall that chain of thinking, as the result of the process of thinking, consists of copies of interrelated knowledge quanta with the same polarization, which can be in any of the nodes of the knowledge tree. In this chain of thinking, the key quantum (concept) is the "mountain"; it contains information about all quanta included in the chain in the form of a list. Each of these quanta has its own set of polarization frequencies, determined by the relationship between their respective concepts, andthe rating value depending on the importance of the linked concept for the utterance as a whole. In the chain of thinking under consideration, "was" has the highest rating, "high" and "overgrown" have lower rating and their ratings are equal, and "very" and "forest" also have the same rating and it is the smallest rating.
The meaning of the whole statement is determined by analyzing the ontological relations between the concepts included in it and their ratings in terms of the value of this statement for the cognitive process. Everyone has his life values, therefore the meaning of the same statement for different people may differ and even change depending on the situation for the same person.

Results
There are a great many entities in the world surrounded us, each of which has a certain set of properties and characteristics that manifest through interaction with other entities. Knowledge about this world, regardless of its origin, generates new knowledge. This is a permanent process, and there are no restrictions on the formation of knowledge structures, except for those that are determined both by a priori forms of thinking and perception and by the nature of the knowledge itself and its emergent ability to self-organize. The mind controls the perception of the surrounding world through the cognitive system and determines the reaction of the organism in each specific situation, but this requires a suitable ontology in which all the acquired empirical knowledge would be stored. A descriptive model of such an ontology in the form of a hierarchical categorical structure called the knowledge tree was proposed in (Moroz, 2020). The nodes of this knowledge tree are containers in which clusters of knowledge (concepts) are accumulated, consisting of quanta of knowledge connected by ontological relations from Table. 1. There is also a detailed description of the procedure for adding a quantum of knowledge to a cluster, controlled by some intelligent device, which results in creating new knowledge and the syntax of which can be considered as a procedural implementation of the process of thinking.
In the paper (Moroz, 2020), only some of the nodes of the knowledge tree were considered in the context of describing the conceptual model of knowledge representation. Below there is a complete description of the nodes of the first three main levels of the knowledge tree that make up its framework. Leaf nodes of the fourth level are not so essential, except for the subtree with the root node "Life of a biological creature", related to Behavioral knowledge, which is of key importance for knowledge organizing. There are a lot of them, and their composition may vary, therefore they are not presented here. Those categories of knowledge that are debatable and may cause questions are provided with appropriate comments.

Discussion
The proposed descriptive ontology of empirical knowledge does not claim to be complete, being only the result of the author's phenomenological research, however, one circumstance is beyond doubtintellect was given to human being by the Creator to cognize the laws of its existence through interaction with the physical world, as well as to realize its place in this world, and, most importantly, to cognize Your Spiritual Self. While cognizing the surrounding world, a person uses both a priori knowledge and a posteriori knowledge, but the transcendental content of a priori forms of thinking does not allow them to be used directly in logical judgments. Nevertheless, one can try to use them as "ontological predicates" following the advice of I. Kant, what was done in this article. A priori categories of reason cannot be formalized, but their regulatory function for logical judgments can be realized in the form of ontological relations between concepts, from which semantic analysis can be built. This paper completes the description of the knowledge representation model presented in (Moroz, 2020), the key idea of which is the primacy of knowledge concerning logical inference and their emergent ability to self-organize.
The offered knowledge representation model in conjunction with the transcendental logic-based ontology of empirical knowledge can be used to create a universal inference engine. The next stage of this research is the software implementation of this model.