Neural networks, genetic algorithms and more ... Myths and reality. Signs
This work is a continuation of everything that was said earlier in the article “ Neural networks, genetic algorithms and more ... Myths and reality. Version II . " In the majority of articles devoted to the analysis of texts that the author was able to study, the analysis of the text refers mainly to two completely practical tasks related to either extracting a context or translating a text from one language to another. In the first case, it is usually a question of either “clearing” the analyzed content and comparing a portion of the text to a standard in accordance with a predetermined taxonomy 1 any entities. For example, parsing addresses, products, etc. In the second case, the search for a correspondence of one block of text written in one language to a block written in another.
Combining both of these options is a statistical, in fact, analysis of sections of contexts, taking into account synonyms of words, well-established expressions. Moreover, the analysis of paths 2 , rhetorical turns and much more goes beyond the scope of this kind of analysis . The reason for this lies in the lack of understanding by modern science even at the philosophical level of some basic issues related to the logic of thinking and decision making, the construction of sign systems underdevelopment of semiotics, etc.
From the point of view of statistical approaches, if you do not take into account the problems with the statement of the problem as such, there is a lack of computational resources for tracking all possible combinations, as well as analyzing semantic structures when implementing various kinds of neural and other combinatorial algorithms. An exception is the analysis of relatively small semantic constructions, such as addresses, goods and services, in the presence of a well-developed reference guide, and even then with not very high indicators of accuracy. Obviously, most fans of neural networks and other weak, inherently AI, may not agree with this postulate. But the practice of using the google.com or Korolev translator Yandex.ru only confirms all of the above. The number of articles on this subject, both scientific and not very large. For example, "Google Translate: A Guide for the Complete Idiot." Some analysis of research on this topic is presented in the article “Comparison of technological approaches to solving problems of data extraction” . The very attractiveness of the same neural networks for solving a certain class of tasks that are not related to the analysis of "sense" is not diminished in any way. The search for statistically significant patterns is a well-developed area.
Let us return to the main question about the reasons why today it is not necessary to talk about any significant success in the analysis of texts. Let us try to outline the main steps that, in the author’s opinion, should be passed before we can talk about the analysis of the text in the full sense of the word.
Analyzing most of the practical solutions in the field of Data mining and Data Science, almost the same picture-model inevitably looms. In this model, ontological directories are built, modules of morphological analysis are added, a base of synonyms for both words and stable expressions is built. Further, based on the accuracy and percentage of the correspondences found in the analyzed text to the elements of one or another anthological reference book, a conclusion is drawn on the effectiveness of one or another algorithm. In this regard, it is not decisive whether the approach built on the basis of neural networks, hash algorithms or any other is chosen. All of them are based on the statistical-frequency principle. As an example, I would like to cite all the frequent publications about how neural networks were able to "self-learn" and "learn" a new meaning. In reality, we can only talk about the “discovery” of no more than new groups or elements of entities within the framework of well-known descriptive models and patterns. To make a decision that a new entity or class is revealed, and not an instance within the framework of a previously described class, such a system cannot even theoretically. The mathematical apparatus does not have the mathematics of induction and abstraction.
Therefore, the existing AI algorithms and libraries are for the most part a dead end in terms of "understanding the meaning" tasks, but, despite this, they will give good results in some areas due to the "simple" extensive growth of marked up knowledge bases, so and computing power. All these victories will lie within the framework of the implementation of Intelligent Agents designed to solve rather narrow practical problems and are quite related to solving the problem of creating an AI-complete mechanism.
In reality, the understanding of the meaning of the text is much more complex and is associated with such concepts as “sign”, “meaning”, “induction”, “value” and much more, as mentioned earlier . This article is devoted to the first of the basic concepts, namely the sign .
Starting to discuss the sign as such, it is necessary to make some clarifications related to the fact that in reality it is necessary to discuss three terms at once, closely related to each other. These are “sign” , “symbol” and “signal” . Regarding terminology, the concept of sign and symbol is often synonymous, the signal is a slightly different entity.
According to Y. Schreider
"it is hardly worth trying to determine exactly what a sign is." But this can be explained from the point of view of the “sign situation”. A sign situation is a pair of sign and signposted. In other words, a sign is what is used as a “substitute” for something. Moreover, in the "real" world there can be neither a sign nor a denoted denotation. Before him, C. Pierce
identified three types of signs:
Pierce defined it as a sign that has a certain natural resemblance to the object to which it refers. That is, its action is based on the actual likeness of the signifier and the signified (for example, the drawing of a lion and the lion himself).
whose action is based on the real adjacency of the signifier and the signified (for example, groan is an index of injury, illness; cheerful, cheerful laughter is an index of joy, happiness). Index signs can also point to an object (pointing with a finger, arrow, shout). The form of these signs is not random for the referent (unlike symbols), but it is not a direct repetition of the referent form (as iconic signs). The form of index signs is connected with the referent by certain relations - for example, causal.
whose action is based on a conditional connection established by agreement between the signified and the signifier (for example, a nod of the head usually means an affirmative answer, however, for some peoples this movement is characterized as a negative answer). Here we see that the connection between the signifier and the signified does not mean a separate object or thing, but the kind of thing and depends on the mentality of a particular culture.
A kind of continuation of this typification is the definition given by Y. Lotman 5. In his understanding, a “symbol” in terms of expression and in terms of content always represents some text, that is, it has some single meaning, closed in itself, and a clearly expressed border, which makes it possible to clearly distinguish him from the surrounding semiotic context. At the same time, further Yu. Lotman, gives an explanation clarifying the position of the “symbol” within the entire totality of signs. The fact that the symbol has the iconic property of “a certain similarity between the planes of expression and content” 6 . Thus, it can be suggested that a “symbol” is a kind of subclass of “signs”, closely related primarily to its presentation in the form of texts or audio recordings, but not a graphic symbol, in other words, only a semantic connection, which is a kind of hint .
The idea that a text is primarily a signal transmitted from one addressee to another was first expressed at the beginning of the last century by C. Morris
. Subsequently, it was remarkably developed, for example, by A. M. Pyatigorsky
in his work, “Some General Remarks Regarding the Consideration of the Text as a Variety of Signal”. Here are some of his points:
“In linguistics, a text is in a certain way limited by the variety of signals ... Another reason in favor of assigning a theory of a text to the science of a signal rather than a language can be the fact that a significant part of the texts (some ideographic hieroglyphic texts, etc.) can be not related to a specific language. ”
“The text is created in a specific, single communication situation - a subjective situation, and is perceived depending on time and place in countless situations.”
Pyatigorsky further writes that the text is inherently a kind of “act of the author’s behavior, in a certain way physically objectified and internally determined by himself, and its creation is such a kind of behavior, the content of which is the description”
“For any act of semiotic awareness, it is essential to highlight significant and insignificant elements in the surrounding reality. "Elements that do not carry values, from the point of view of the modeling system, as it were, do not exist." 9
Thus, any flow of information and text, according to a number of authors, should be considered, first of all, as a signal that should cause any reaction in the recipient.
Context. According to the apt expression of the same Yu. Lotman, each text has its own “audience”, which has some common memory. Thus, the task of text analysis is further complicated by the fact that the “analyzer” must correctly predict the target group of text consumers.
In practice, the situation is even “worse”, according to some authors, any text is built on the basis of “images-models that have a syncretic
verbal-visual existence” 11 , including the mixing and clash of contexts in order to rhetorically enhance meaning. In the case of texts intended for a wide audience, the task of choosing the right context and a set of codes can be partially solved by analyzing terminology, established expressions, etc. In the case of private correspondence, the search for the correct semiotic field is an almost unsolvable task that requires an in-depth analysis of the history of the correspondent’s relations, the construction of their own code structure, including paths, “their own”, invented words, etc.
Page ктура знаковых систем
Page ктуру знаковых систем представляет собой многогранное и многоуровневое явление. The sciences that can be described as sections of semiotics include syntactics, semantics, pragmatics, linguistics, rhetoric, cultural studies, and a number of others. Interdisciplinary sciences affecting the sphere of interests of semiotics include cultural anthropology, ethnosemiotics and ethnography, history, philosophy, mathematics, etc.
At the lower level, we can talk about the structure of the sign, on the average, about semantic and semiotic structures. At the highest level - about the semiosphere as a whole.
For the first time G. Frege, C. Ogden and A. Richards 12 , De Saussure 13 , C. Morris and several other authors spoke about the structure of the sign back in the 1920s. There are a huge number of works on this subject, for example, the work of Schrader 14 , or articles in the Russian Journal and many others, which do not make sense to repeat within the framework of this article. We dwell briefly only on the Frege triangle:
All these works are united by the analysis of the relationship of sign-concept-denotation. Where the concept is the information that the sign carries about possible denotations 15 , and denotation is what is indicated. C. Morris identifies an additional abstract concept - designate. In his understanding, a designation is not a thing, but a kind of object or a class of objects, and a class can include either many members, or only one member, or no members at all 16 . The concept of the sign is described above. Thus, any text can be considered in two ways: on the one hand, it is a kind of superposition of signs, and on the other - their concepts, in some cases the whole text will be a sign entirely.
Everything is much more ambiguous with the definition of sign-to-signal relationships. As mentioned above, in the works of such researchers as C. Morris, A. Pyatigorsky, R. Jacobson, one of the priority messages is the fact that any text should first be perceived as a signal. If at the beginning of the last century this was given great importance, today, for the sake of data translation and extraction, the lion's share of research is devoted to syntagmatic and semantic analysis, the construction of various kinds of thesauruses.
All that is mentioned today in connection with signals is that, some of the signs can be attributed to signals or, in Pierce's terminology, to indices. In reality, the picture is completely opposite - signs are part of the signal system. If we carefully consider the analysis of the sign-signal pair, then we move on to a completely different field of research related to additional signal properties with respect to the sign.
The main features of the signal include the following:
It has a physical entity in the form of a distribution channel.
According to the concept proposed by A. Pyatigorsky, the text has such basic characteristics as spatiality, time, and objectivity. Spatiality is understood as the very fact of the presence of an information carrier, as its form. Form can be significant. For example, writing and newspaper clipping have completely different sensitivity to the assessment and interpretation of the meaning of the text.
The channel is used to transmit non-verbal and verbal characters.
Most often, speaking of understanding the meaning, we operate on the tasks of analyzing texts or recognizing emotions on a person’s face. These tasks essentially boil down to two fairly isolated channels - texts and images.
Waiting for feedback - reaction is supposed.
The reason for the appearance of any signal is the need or the need to receive any reaction. It may be aimed at establishing feedback, but may also be one-sided. But the presence of any reaction is a prerequisite.
- The accuracy of the reaction may vary.
Varying the accuracy of the reaction is a function that depends on such parameters as the presence of noise, the presence of a general multiplicity of concepts, associative series. All this can be expressed by a generalized concept of a common (identical) previous experience. In this case, noise should be understood not as the truth or falsity of the flow of signs, but rather its mutual consistency of signs. This problem is exacerbated by the presence of pathways and rhetorical rhetoric in general, which is not an outlier context. They serve to enhance the flow of signals.
signal strength and quality;
The property of the strength and quality of the impact depends on many factors, but mainly on the availability of general experience, the state of the addressee at the time of receipt of the signal, and the totality of signs by which the signal is transmitted. In addition, one should take into account the “density” and speed of the transmitted stream of signs that limit the limit of perception, and the strength of the emotional impact that limits the possibilities of the psyche, in the case of the target person.
The theme of emotions in itself is limitless. As part of the objectives of the current study, two questions are of interest. Is emotion an independent signal? Does it in itself have a symbolic nature, i.e. Is there a concept and designation? Is emotion a sign property or one of the characteristics of a signaling channel? Due to the limitations of the format of the article, we confine ourselves to the statement that emotion is at least a full-fledged signal with an unobvious symbolic nature. Emotional coloring is a characteristic of a channel that changes in time as part of the channel’s process.
Part of the above features in one way or another have something in common with the well-known theory of signals in the theory of information.
Semiosphere - habitat signs
The term semiosphere
was introduced by J. Lotman, by analogy with the biosphere of V. Vernadsky, as a common cultural space, including the
language as part of it.
The semiotic system, according to Lotman, has binary, asymmetry and heterogeneity. The principle of binary assumes that any language (semiotic system) sooner or later undergoes fragmentation into other languages (systems). In this case, it is necessary to take into account the presence of the reverse mechanism - unification. As a result, new languages of art appear on the one hand, and new canons arise on the other. The principle of asymmetry suggests that such crushing occurs in different proportions. The principle of heterogeneity suggests that within the framework of the semiosphere, structures of completely different nature can exist. Simplistically, various memes, local concepts about the truth of certain statements, various meanings of the same concepts and denotations, and much more can act as such. В результате могут возникать ситуации как частичной, так и полной непереводимости.
Based on the foregoing, the conclusion suggests itself that there is no single semiotic structure or any isolated systems. In contrast, it should be said
that there is a global semiosphere, which is a kind of “superposition” of various semiotic systems.
The construction of a model of a communication channel is one of the most important tasks, without the solution of which it is impossible to analyze the processes during which there is an exchange of signals, and signs, in particular.
Later, other essentially extended models were proposed. For example, the “theater model” by Y. Kristeva 20 : “author-addressee-others-situation”.
All these models indicate that in itself any of the models, in addition to operating with the general concepts of the addressee and the recipient, is highly dependent on such concepts as “context”. According to Y. Kristeva 21 , one of the basic problems in the analysis of the transmitted content is the lack of a single meaning, and therefore the rejection of the very possibility of an unambiguous interpretation. The distortion or differences in the interpretations of meaning are caused by the denial of the right to the subject-author of “sacredness”, “magic” and “effect”.
By "sacredness" is meant the absence in the right to the sole meaning. Under "magic" - the inability to protect the subject from the dominant semiosphere, i.e. from possible changes in interpretations in certain semiotic systems.
From the point of view of the topic of the article, the presence of these models indicates the fact that it will be necessary to design not only the channel itself, but also implement, for example, such subsystems as:
- network base-graphs of signs, which themselves can be encapsulated into each other;
- groups of subjects (recipients and recipients)
- Network models of possible contexts.
Omni Channel 22
Most of the works related to semiotics and semantics operate with such concepts as text or image isolated from the point of view of analysis. But without a study of the effects, which imposes the presence of a large number of parallel and mutually integrated flows, it is not necessary to talk about a full-fledged AI, even in tasks of “simple” text analysis. When transmitting information, for example, educational films, lecture courses, etc. at the same time there are both text and audio-video streams of information.
Until today, the author has not come across how much significant work on this topic.
This topic is a huge layer of issues, a detailed discussion of which is not possible in a separate article. Therefore, the main task was to show the main blocks, without the implementation of which further promotion is simply impossible.
Existing solutions, of those that are known to the author at the moment, either completely ignore the problem of signs, or have a greatly simplified approach to the problem of analysis and operation of signs.
For further advancement, it is necessary to solve the following engineering problems:
- Creation of unified complexes of bases and operating rules, combining various classes of sign systems such as text, graphic, video and audio signs. These complexes should include such subsystems as signs, concepts and denotations in the context of cultural, professional, historical and several other dimensions.
- Creation of bases and types of recipient entities, taking into account the measurements specified in the previous paragraph.
- Creation of bases of interconnections of semiotic structures, and not just blocks of interconnected texts and various kinds of thesauruses.
- The creation of bases and rules for the generation and operation of the overall emotional impact of signs.
- The created system must necessarily take into account the omnichannel problem, even in the case of operating within only one of the channels.
- Refusal of attempts to obtain a single meaning in favor of constructing rather probabilistic results characteristic of a particular semiotic structure, the interpretation of which cumulatively depends on previous contexts analyzed, including related ones.
- The concept of fidelity and falsity of statements, formed by a superposition of statements formed by signs, is a certainty only in the context of one moment in time. Therefore, subsystems for monitoring and analysis of such superpositions should be implemented. Models have been developed that correct the main flow of conclusions.
1 TAXONOMY - the science of classifying complex objects of reality (wildlife, Earth's structure, ethnic communities, language, etc.) Ozhegov's Explanatory Dictionary. S.I. Ozhegov, N.Yu. Shvedova. 1949-1992.
2 TROPES (from the Greek tropos - turn, turn of speech), 1) in the style and poetics the use of the word in the figurative sense, in which there is a shift in the semantics of the word from its direct meaning to the figurative. Three types of pathways are built on the relationship between the direct and figurative meanings of the word: the relationship by similarity ( metaphor ), by contrast ( oxymoron ), by adjacency ( metonymy ). Various ways of word transformations (from word to image) are an important element of artistic thinking. Along with the main types of paths, their varieties are also distinguished ( irony , synecdoch, hyperbole , litota , epithet , etc.). 2) (Musical) in the Gregorian chorus, the insertion (or series of insertions) into the canonical hymn is textual or textless. The modern encyclopedia. 2000.
3 Schreider Yu.A. The Logic of Sign Systems: Elements of Semiotics. Book House "LIBRICOM", 2012. Page 4
5 Lotman Yu.M. Inside the thinking worlds. Издательство Азбука, 2015. С.158
6 Lotman Yu.M. Inside the thinking worlds. Publishing House ABC, 2015.
P.171 7 Ch.Morris Fundations of the theory of Sings. Chicago, 1938
8 A.M. Пятигорский "Некоторые общие замечания относительно рассмотрения текста как разновидности сигнала». Page ктурно-типологические исследования. Сборник статей. М.: Издательство Академии Наук СССР, 1962 Page 144-154
9 Lotman Yu.M. Inside the thinking worlds. Издательство Азбука, 2015. С.90
10 Синкретический — филос. состоящий из разнородных элементов, но являющийся целостным.
11 Lotman Yu.M. Inside the thinking worlds. Издательство Азбука, 2015. С.127
12 C. K. Ogden and I. A. Richards, The Meaning of Meaning subtitled A Study of the Influence of Language upon Thought and of the Science of Symbolism, 1923, Magdalene College, University of Cambridge, p. 11
13 Saussure Ferdinand de, Course in General Linguistics: Published by S. Balli and A. Séche with the participation of A. Riedlinger. Per.s FR / Ed. And with a note. R.O. Shor., Book House "LIBRICOM", 2016.
14 Schreider Yu.A. The Logic of Sign Systems: Elements of Semiotics. Book House "LIBRICOM", 2012.
15 Schreider Yu.A. The Logic of Sign Systems: Elements of Semiotics. Book House "LIBRICOM", 2012. S.10.
16 Ch.Morris Fundations of the theory of Sings. Chicago, 1938 г.
17 Lotman Yu.M. Inside the thinking worlds. Publishing House ABC, 2015. P.176
18 Jakobson R. Lingustics and Poetics. Style in Linguage. Cambridge, 1964. P. 353
19 Lotman Yu.M. Inside the thinking worlds. Publishing house ABC, 2015. P.30
20 Kristeva Yu. Semiotika: Studies on semanalysis. Academic project, 2015.
21 Kristeva Yu. Semiotics: Studies on semanalysis. Academic project, 2015. C 13.
22 The term "omnichannel" was introduced by the author by analogy with the concept of omnichannel adopted in marketing.