Yair Neuman (b. 1968) is a polymath whose expertise is in interdisciplinary research where he draws on diverse disciplines to creatively address real world and academic challenges.
He is the head of the Functor Lab, at BGU.
Prof. Neuman has published numerous papers and academic books and was a visiting scholar/Prof. at M.I.T, University of Toronto, University of Oxford, and Weizmann Institute of Science.
Beyond his academic work, he developed state-of-the-art algorithms for social and cognitive computing, such as those he developed for IARPA, and is currently developing for a DARPA project.
Context is necessary for understanding human behavior. However, so far, the concept of context ha... more Context is necessary for understanding human behavior. However, so far, the concept of context has mostly been treated in a way that lacks any clear relevance for using, developing and engineering intelligent systems. In this book, the author explains the importance of context for understanding human behavior, presents a theory of context, and shows how AI, specifically Large Language Models such as GPT, can support our understanding of context when analyzing human behavior as expressed in texts ranging from conversations to short stories.
Drawing on years of R&D and academic publications in top rated journals, the author provides the reader with a simple and deep understanding of context and its modeling for specific challenges, from identifying social norm violations to understanding conversations going awry and stories by great authors. The book may interest a wide variety of readers seeking to incorporate AI into their understanding of human behavior.
Crowds are misleading, both in their simplicity and in their complexity. On the one hand, they be... more Crowds are misleading, both in their simplicity and in their complexity. On the one hand, they behave according to expected trends; on the other, they present sudden shifts and frantic, unexpected behavior. Therefore, “betting against the crowd,” whether in politics, sports, or finance, requires a deep understanding of the crowd’s dynamics. In this book, Prof. Neuman addresses this challenge by delving into the complexity of crowds. The book involves foundational issues and novel ideas, such as why crowds behave unexpectedly, why betting against the crowd is possible only in short time frames, why is it important to be attentive to suspicious signs that are indicative of the crowd’s behavior, and why the long tail of fatalities in armed conflicts leaves us surprised by blitz attacks of violent mobs. The book combines scientific knowledge, experiments, and accessible, often humorous, exposition. It can be read by anyone with a basic science education who seeks to understand crowds and how one can act within and against them.
This concise volume offers an accessible introduction to state-of-the-art artificial intelligence... more This concise volume offers an accessible introduction to state-of-the-art artificial intelligence (AI) language models, providing a platform for their use in textual interpretation across the humanities and social sciences.
The book outlines the affordances of new technologies for textual analysis, which has historically employed established approaches within the humanities. Neuman, Danesi, and Vilenchik argue that these different forms of analysis are indeed complementary, demonstrating the ways in which AI-based perspectives echo similar theoretical and methodological currents in traditional approaches while also offering new directions for research. The volume showcases examples from a wide range of texts, including novels, television shows, and films to illustrate the ways in which the latest AI technologies can be used for "dialoguing" with textual characters and examining textual meaning coherence.
Illuminating the potential of AI language models to both enhance and extend research on the interpretation of texts, this book will appeal to scholars interested in cognitive approaches to the humanities in such fields as literary studies, discourse analysis, media studies, film studies, psychology, and artificial intelligence.
Searching for a needle in a haystack is an important task in several contexts of data analysis and decision-making. Examples include identifying the insider threat within an organization, the prediction of failure in industrial production, or pinpointing the unique signature of a solo perpetrator, such as a school shooter or a lone wolf terrorist. It is a challenge different from that of identifying a rare event (e.g., a tsunami) or detecting anomalies because the "needle" is not easily distinguished from the haystack. This challenging context is imbued with particular difficulties, from the lack of sufficient data to train a machine learning model through the identification of the relevant features and up to the painful price of false alarms, which might cause us to question the relevance of machine learning solutions even if they perform well according to common performance criteria. In this book, Prof. Neuman approaches the problem of finding the needle by specifically focusing on the human factor, from solo perpetrators to insider threats. Providing for the first time a deep, critical, multidimensional, and methodological analysis of the challenge, the book offers data scientists and decision makers a deep scientific foundational approach combined with a pragmatic practical approach that may guide them in searching for a needle in a haystack.
Most of us are intuitively familiar with small social systems, such as families and soccer teams.... more Most of us are intuitively familiar with small social systems, such as families and soccer teams. Surprisingly, though, most of us are unaware of how complex these systems are or of the fact that they have a unique character distinguishing them from both populations and individuals. The current manuscript, which emerged from high-level scientific publications on the subject, aims to bridge this gap in our understanding of small social systems. The book aims to explain, illustrate, and model the unique and fascinating nature of small (social) systems by relying on deep scientific foundations and by using examples from sport, movies, music, and the martial arts. To support its friendly exposition of challenging scientific ideas, the book also discusses entertaining questions such as (1) why inviting your mother-in-law to dinner might be a challenging event, for reasons you have never considered; (2) why soccer teams should be messy in order to win; (3) why Nazis are deeply wrong in their understanding of the importance of entropy; and (4) why “panda fighters” failed in the UFC (Ultimate Fighting Championship).
The old practices of interpretation have been exhausted, and the humanities and social sciences a... more The old practices of interpretation have been exhausted, and the humanities and social sciences are facing a crisis. Is there a way out of the labyrinth of reading? In this book, Professor Neuman presents a challenging approach to interpreting texts and reading literature through the spectacles of conceptual mathematics. This approach strives to avoid the simplicity of a quantitative approach to the analysis of literature as well as both the relativistic and the ideological dangers facing a qualitative reading of a text. The approach is introduced in a rigorous and accessible manner and woven with insights gained from various fields. Taking us on a challenging journey from Ovid's Metamorphoses to Nick Cave's The Death of Bunny Munro, the book shows how we may gain a deeper understanding of literature and the aesthetic experience of reading.
"Conceptual Mathematics and Literature employs an interdisciplinary mathematical approach that is a uniquely insightful interpretive analysis of literary works. It offers novel ways of comprehending previously unnoticed underlying patterns of meaning. Numerous figures illustrate and reinforce the author's brilliant and illuminating mode of critical inquiry."-Frank Nuessel, Professor of Modern Languages and Linguistics, University Scholar at the University of Louisville
Mathematical Structures of Natural Intelligence uncovers mathematical structures underlying natur... more Mathematical Structures of Natural Intelligence uncovers mathematical structures underlying natural intelligence and applies category theory as a modeling language for understanding human cognition, giving readers new insights into the nature of human thought. In this context, the book explores various topics and questions, such as the human representation of the number system, why our counting ability is different from that which is evident among non-human organisms, and why the idea of zero is so difficult to grasp. The book is organized into three parts: the first introduces the general reason for studying general structures underlying the human mind; the second part introduces category theory as a modeling language and use it for exposing the deep and fascinating structures underlying human cognition; and the third applies the general principles and ideas of the fi st two parts to reaching a better understanding of challenging aspects of the human mind such as our understanding of the number system, the metaphorical nature of our thinking and the logic of our unconscious dynamics.
This is the cover page of my forthcoming book published by Springer. The book introduces the fiel... more This is the cover page of my forthcoming book published by Springer. The book introduces the field of computational personality analysis.
"This book presents programmatic texts on biosemiotics, written collectively by the theoreticians... more "This book presents programmatic texts on biosemiotics, written collectively by the theoreticians in the field (Deacon, Emmeche, Favareau, Hoffmeyer, Kull, Markoš, Pattee, Stjernfelt). In addition, the book includes chapters which focus closely on semiotic case studies (Bruni, Kotov, Maran, Neuman, Turovski).
According to the central thesis of biosemiotics, sign processes characterise all living systems and the very nature of life, and their diverse phenomena can be best explained via the dynamics and typology of sign relations. The authors are therefore presenting a deeper view on biological evolution, intentionality of organisms, the role of communication in the living world and the nature of sign systems — all topics which are described in this volume. This has important consequences on the methodology and epistemology of biology and study of life phenomena in general, which the authors aim to help the reader better understand."
The computational analysis of human personality has mainly focused on the Big Five personality th... more The computational analysis of human personality has mainly focused on the Big Five personality theory, and the psychodynamic approach is almost nonexistent despite its rich theoretical grounding and relevance to various tasks. Here, we provide a data set of 4972 synthetic utterances corresponding with five personality dimensions described by the psychodynamic approach: depressive, obsessive, paranoid, narcissistic, and anti-social psychopathic. The utterances have been generated through AI with a deep theoretical orientation that motivated the design of prompts for GPT-4. The dataset has been validated through 14 tests, and it may be relevant for the computational study of human personality and the design of authentic persona in digital domains, from gaming to the artistic generation of movie characters.
The violation of social norms in TV and cinema is a well-known source of humor and catharsis, and... more The violation of social norms in TV and cinema is a well-known source of humor and catharsis, and researchers in digital humanities may benefit from the automatic identification of social norm violations. In this article, we introduce a novel methodology for identifying and analyzing the violation of social norms in textual data and illustrate it in the analysis of movie plots. The methodology leans on zero-shot classification, specifically relevant when massive, labeled datasets are unavailable. We test our methodology and provide researchers with (1) a theoretically grounded tool for screening textual data for social norm violation and with new datasets that include (2) 6,806 embarrassing situations from movie plots and their hypothesized violated norm and (3) 3,059 movie plots with their average embarrassment score.
Herd behavior is a powerful source of growth in financial markets. However, as available energy r... more Herd behavior is a powerful source of growth in financial markets. However, as available energy resources limit exponential growth, we should expect periods where an upward trend is balanced toward equilibrium or reverse its direction toward decline. This paper proposes a novel approach for modeling herd behavior and predicting a trend reversal in financial markets. Our approach relies on two key metrics: asymmetry and 'steps to symmetry.' We use Machine Learning to identify hidden patterns in the fluctuations of these metrics and use the patterns for predicting a transition from exponential growth. Analyzing three datasets of stock prices, we present solid empirical evidence supporting the proposed approach.
It has been realized that situational dimensions, as represented by human beings, are crucial for... more It has been realized that situational dimensions, as represented by human beings, are crucial for understanding human behavior. The Riverside Situational Q (RSQ) is a tool that measures the psychological properties of situations. However, the RSQ-4 includes only 90 items and may have limited use for researchers interested in measuring situational dimensions using a computational approach. Here we present a corpus of 10,000 artificially generated situations corresponding mostly with the RSQ-4. The dataset was generated using GPT, the state-of-the-art large language model. The dataset validity is established through inter-judge reliability, and four experiments on large datasets support its quality. The dataset and the code used for generating 100 situational dimensions may be useful for researchers interested in measuring situational dimensions in textual data.
Identifying social norms and their violation is a challenge facing several projects in computatio... more Identifying social norms and their violation is a challenge facing several projects in computational science. This paper presents a novel approach to identifying social norm violations. We used GPT-3, zero-shot classification, and automatic rule discovery to develop simple predictive models grounded in psychological knowledge. Tested on two massive datasets, the models present significant predictive performance and show that even complex social situations can be functionally analyzed through modern computational tools. The APA Dictionary of Psychology defines social norms 1 as socially determined standards that indicate typical and proper behaviors in a given social context. Some norms are universal (e.g., it is wrong to murder), some are contextual within the culture (e.g., it is not legitimate to expose your body in public, but it is legitimate to expose it on a nudist beach), and some are highly particular (e.g., in the United States it is expected to tip bartenders). Moreover, the dictionary suggests that, whether implicitly or implicitly, the norms proscribe actions that should be avoided as they violate a social norm (i.e., social norm violation). Although some basic social norms seem universal, they are expressed with cultural variations 2 , as sometimes observed in cross-cultural encounters, such as those that appear in the black comedy mockumentary Borat 3. While social norms and their violations have been intensively studied in psychology and the social sciences 4,5 the automatic identification of social norms and their violation is an open challenge that may be highly important for several projects, such as the engineering of digital interpreters supporting cross-cultural interactions and DAPRA's computational cultural understanding program 6 that funded the current project. It is an open challenge because we first have to identify the features/signals/variables indicating that a social norm has been violated. So far, the operational definition of theoretically grounded features is a task that has not been accomplished. For example, arriving at your office drunk and dirty is a violation of a social norm among the majority of working people. However, "teaching" the machine/computer that such behavior is a norm violation is far from trivial. For classifying behavior as violating a social norm, the machine should learn that certain measurable variables/ features signal that a norm has been violated. In this sense, our paper aims to bridge the gap between social sciences and data science by engineering simple and theoretically grounded models that can successfully classify cases that involve norm violation. This paper presents a novel approach to identifying social norm violations. The models developed in this paper rely on state-of-the-art language models such as GPT3 7,8 NLI-based Zero Shot Text Classification 9,10 and automatic rule discovery 11 , but are also grounded in deep theoretical understanding of human psychology and social emotions. It must be explained in a nutshell that we
Human interlocutors may use emotions as an important signaling device for coordinating an interac... more Human interlocutors may use emotions as an important signaling device for coordinating an interaction. In this context, predicting a significant change in a speaker’s emotion may be important for regulating the interaction. Given the nonlinear and noisy nature of human conversations and relatively short time series they produce, such a predictive model is an open challenge, both for modeling human behavior and in engineering artificial intelligence systems for predicting change. In this paper, we present simple and theoretically grounded models for predicting the direction of change in emotion during conversation. We tested our approach on textual data from several massive conversations corpora and two different cultures: Chinese (Mandarin) and American (English). The results converge in suggesting that change in emotion may be successfully predicted, even with regard to very short, nonlinear, and noisy interactions.
Revenge is a powerful motivating force reported to underlie the behavior of various solo perpetra... more Revenge is a powerful motivating force reported to underlie the behavior of various solo perpetrators, from school shooters to right wing terrorists. In this paper, we develop an automated methodology for identifying vengeful themes in textual data. Testing the model on four datasets (vengeful texts from social media, school shooters, Right Wing terrorist and Islamic terrorists), we present promising results, even when the methodology is tested on extremely imbalanced datasets. The paper not only presents a simple and powerful methodology that may be used for the screening of solo perpetrators but also validate the simple theoretical model of revenge. https://arxiv.org/abs/2205.01731
Context is necessary for understanding human behavior. However, so far, the concept of context ha... more Context is necessary for understanding human behavior. However, so far, the concept of context has mostly been treated in a way that lacks any clear relevance for using, developing and engineering intelligent systems. In this book, the author explains the importance of context for understanding human behavior, presents a theory of context, and shows how AI, specifically Large Language Models such as GPT, can support our understanding of context when analyzing human behavior as expressed in texts ranging from conversations to short stories.
Drawing on years of R&D and academic publications in top rated journals, the author provides the reader with a simple and deep understanding of context and its modeling for specific challenges, from identifying social norm violations to understanding conversations going awry and stories by great authors. The book may interest a wide variety of readers seeking to incorporate AI into their understanding of human behavior.
Crowds are misleading, both in their simplicity and in their complexity. On the one hand, they be... more Crowds are misleading, both in their simplicity and in their complexity. On the one hand, they behave according to expected trends; on the other, they present sudden shifts and frantic, unexpected behavior. Therefore, “betting against the crowd,” whether in politics, sports, or finance, requires a deep understanding of the crowd’s dynamics. In this book, Prof. Neuman addresses this challenge by delving into the complexity of crowds. The book involves foundational issues and novel ideas, such as why crowds behave unexpectedly, why betting against the crowd is possible only in short time frames, why is it important to be attentive to suspicious signs that are indicative of the crowd’s behavior, and why the long tail of fatalities in armed conflicts leaves us surprised by blitz attacks of violent mobs. The book combines scientific knowledge, experiments, and accessible, often humorous, exposition. It can be read by anyone with a basic science education who seeks to understand crowds and how one can act within and against them.
This concise volume offers an accessible introduction to state-of-the-art artificial intelligence... more This concise volume offers an accessible introduction to state-of-the-art artificial intelligence (AI) language models, providing a platform for their use in textual interpretation across the humanities and social sciences.
The book outlines the affordances of new technologies for textual analysis, which has historically employed established approaches within the humanities. Neuman, Danesi, and Vilenchik argue that these different forms of analysis are indeed complementary, demonstrating the ways in which AI-based perspectives echo similar theoretical and methodological currents in traditional approaches while also offering new directions for research. The volume showcases examples from a wide range of texts, including novels, television shows, and films to illustrate the ways in which the latest AI technologies can be used for "dialoguing" with textual characters and examining textual meaning coherence.
Illuminating the potential of AI language models to both enhance and extend research on the interpretation of texts, this book will appeal to scholars interested in cognitive approaches to the humanities in such fields as literary studies, discourse analysis, media studies, film studies, psychology, and artificial intelligence.
Searching for a needle in a haystack is an important task in several contexts of data analysis and decision-making. Examples include identifying the insider threat within an organization, the prediction of failure in industrial production, or pinpointing the unique signature of a solo perpetrator, such as a school shooter or a lone wolf terrorist. It is a challenge different from that of identifying a rare event (e.g., a tsunami) or detecting anomalies because the "needle" is not easily distinguished from the haystack. This challenging context is imbued with particular difficulties, from the lack of sufficient data to train a machine learning model through the identification of the relevant features and up to the painful price of false alarms, which might cause us to question the relevance of machine learning solutions even if they perform well according to common performance criteria. In this book, Prof. Neuman approaches the problem of finding the needle by specifically focusing on the human factor, from solo perpetrators to insider threats. Providing for the first time a deep, critical, multidimensional, and methodological analysis of the challenge, the book offers data scientists and decision makers a deep scientific foundational approach combined with a pragmatic practical approach that may guide them in searching for a needle in a haystack.
Most of us are intuitively familiar with small social systems, such as families and soccer teams.... more Most of us are intuitively familiar with small social systems, such as families and soccer teams. Surprisingly, though, most of us are unaware of how complex these systems are or of the fact that they have a unique character distinguishing them from both populations and individuals. The current manuscript, which emerged from high-level scientific publications on the subject, aims to bridge this gap in our understanding of small social systems. The book aims to explain, illustrate, and model the unique and fascinating nature of small (social) systems by relying on deep scientific foundations and by using examples from sport, movies, music, and the martial arts. To support its friendly exposition of challenging scientific ideas, the book also discusses entertaining questions such as (1) why inviting your mother-in-law to dinner might be a challenging event, for reasons you have never considered; (2) why soccer teams should be messy in order to win; (3) why Nazis are deeply wrong in their understanding of the importance of entropy; and (4) why “panda fighters” failed in the UFC (Ultimate Fighting Championship).
The old practices of interpretation have been exhausted, and the humanities and social sciences a... more The old practices of interpretation have been exhausted, and the humanities and social sciences are facing a crisis. Is there a way out of the labyrinth of reading? In this book, Professor Neuman presents a challenging approach to interpreting texts and reading literature through the spectacles of conceptual mathematics. This approach strives to avoid the simplicity of a quantitative approach to the analysis of literature as well as both the relativistic and the ideological dangers facing a qualitative reading of a text. The approach is introduced in a rigorous and accessible manner and woven with insights gained from various fields. Taking us on a challenging journey from Ovid's Metamorphoses to Nick Cave's The Death of Bunny Munro, the book shows how we may gain a deeper understanding of literature and the aesthetic experience of reading.
"Conceptual Mathematics and Literature employs an interdisciplinary mathematical approach that is a uniquely insightful interpretive analysis of literary works. It offers novel ways of comprehending previously unnoticed underlying patterns of meaning. Numerous figures illustrate and reinforce the author's brilliant and illuminating mode of critical inquiry."-Frank Nuessel, Professor of Modern Languages and Linguistics, University Scholar at the University of Louisville
Mathematical Structures of Natural Intelligence uncovers mathematical structures underlying natur... more Mathematical Structures of Natural Intelligence uncovers mathematical structures underlying natural intelligence and applies category theory as a modeling language for understanding human cognition, giving readers new insights into the nature of human thought. In this context, the book explores various topics and questions, such as the human representation of the number system, why our counting ability is different from that which is evident among non-human organisms, and why the idea of zero is so difficult to grasp. The book is organized into three parts: the first introduces the general reason for studying general structures underlying the human mind; the second part introduces category theory as a modeling language and use it for exposing the deep and fascinating structures underlying human cognition; and the third applies the general principles and ideas of the fi st two parts to reaching a better understanding of challenging aspects of the human mind such as our understanding of the number system, the metaphorical nature of our thinking and the logic of our unconscious dynamics.
This is the cover page of my forthcoming book published by Springer. The book introduces the fiel... more This is the cover page of my forthcoming book published by Springer. The book introduces the field of computational personality analysis.
"This book presents programmatic texts on biosemiotics, written collectively by the theoreticians... more "This book presents programmatic texts on biosemiotics, written collectively by the theoreticians in the field (Deacon, Emmeche, Favareau, Hoffmeyer, Kull, Markoš, Pattee, Stjernfelt). In addition, the book includes chapters which focus closely on semiotic case studies (Bruni, Kotov, Maran, Neuman, Turovski).
According to the central thesis of biosemiotics, sign processes characterise all living systems and the very nature of life, and their diverse phenomena can be best explained via the dynamics and typology of sign relations. The authors are therefore presenting a deeper view on biological evolution, intentionality of organisms, the role of communication in the living world and the nature of sign systems — all topics which are described in this volume. This has important consequences on the methodology and epistemology of biology and study of life phenomena in general, which the authors aim to help the reader better understand."
The computational analysis of human personality has mainly focused on the Big Five personality th... more The computational analysis of human personality has mainly focused on the Big Five personality theory, and the psychodynamic approach is almost nonexistent despite its rich theoretical grounding and relevance to various tasks. Here, we provide a data set of 4972 synthetic utterances corresponding with five personality dimensions described by the psychodynamic approach: depressive, obsessive, paranoid, narcissistic, and anti-social psychopathic. The utterances have been generated through AI with a deep theoretical orientation that motivated the design of prompts for GPT-4. The dataset has been validated through 14 tests, and it may be relevant for the computational study of human personality and the design of authentic persona in digital domains, from gaming to the artistic generation of movie characters.
The violation of social norms in TV and cinema is a well-known source of humor and catharsis, and... more The violation of social norms in TV and cinema is a well-known source of humor and catharsis, and researchers in digital humanities may benefit from the automatic identification of social norm violations. In this article, we introduce a novel methodology for identifying and analyzing the violation of social norms in textual data and illustrate it in the analysis of movie plots. The methodology leans on zero-shot classification, specifically relevant when massive, labeled datasets are unavailable. We test our methodology and provide researchers with (1) a theoretically grounded tool for screening textual data for social norm violation and with new datasets that include (2) 6,806 embarrassing situations from movie plots and their hypothesized violated norm and (3) 3,059 movie plots with their average embarrassment score.
Herd behavior is a powerful source of growth in financial markets. However, as available energy r... more Herd behavior is a powerful source of growth in financial markets. However, as available energy resources limit exponential growth, we should expect periods where an upward trend is balanced toward equilibrium or reverse its direction toward decline. This paper proposes a novel approach for modeling herd behavior and predicting a trend reversal in financial markets. Our approach relies on two key metrics: asymmetry and 'steps to symmetry.' We use Machine Learning to identify hidden patterns in the fluctuations of these metrics and use the patterns for predicting a transition from exponential growth. Analyzing three datasets of stock prices, we present solid empirical evidence supporting the proposed approach.
It has been realized that situational dimensions, as represented by human beings, are crucial for... more It has been realized that situational dimensions, as represented by human beings, are crucial for understanding human behavior. The Riverside Situational Q (RSQ) is a tool that measures the psychological properties of situations. However, the RSQ-4 includes only 90 items and may have limited use for researchers interested in measuring situational dimensions using a computational approach. Here we present a corpus of 10,000 artificially generated situations corresponding mostly with the RSQ-4. The dataset was generated using GPT, the state-of-the-art large language model. The dataset validity is established through inter-judge reliability, and four experiments on large datasets support its quality. The dataset and the code used for generating 100 situational dimensions may be useful for researchers interested in measuring situational dimensions in textual data.
Identifying social norms and their violation is a challenge facing several projects in computatio... more Identifying social norms and their violation is a challenge facing several projects in computational science. This paper presents a novel approach to identifying social norm violations. We used GPT-3, zero-shot classification, and automatic rule discovery to develop simple predictive models grounded in psychological knowledge. Tested on two massive datasets, the models present significant predictive performance and show that even complex social situations can be functionally analyzed through modern computational tools. The APA Dictionary of Psychology defines social norms 1 as socially determined standards that indicate typical and proper behaviors in a given social context. Some norms are universal (e.g., it is wrong to murder), some are contextual within the culture (e.g., it is not legitimate to expose your body in public, but it is legitimate to expose it on a nudist beach), and some are highly particular (e.g., in the United States it is expected to tip bartenders). Moreover, the dictionary suggests that, whether implicitly or implicitly, the norms proscribe actions that should be avoided as they violate a social norm (i.e., social norm violation). Although some basic social norms seem universal, they are expressed with cultural variations 2 , as sometimes observed in cross-cultural encounters, such as those that appear in the black comedy mockumentary Borat 3. While social norms and their violations have been intensively studied in psychology and the social sciences 4,5 the automatic identification of social norms and their violation is an open challenge that may be highly important for several projects, such as the engineering of digital interpreters supporting cross-cultural interactions and DAPRA's computational cultural understanding program 6 that funded the current project. It is an open challenge because we first have to identify the features/signals/variables indicating that a social norm has been violated. So far, the operational definition of theoretically grounded features is a task that has not been accomplished. For example, arriving at your office drunk and dirty is a violation of a social norm among the majority of working people. However, "teaching" the machine/computer that such behavior is a norm violation is far from trivial. For classifying behavior as violating a social norm, the machine should learn that certain measurable variables/ features signal that a norm has been violated. In this sense, our paper aims to bridge the gap between social sciences and data science by engineering simple and theoretically grounded models that can successfully classify cases that involve norm violation. This paper presents a novel approach to identifying social norm violations. The models developed in this paper rely on state-of-the-art language models such as GPT3 7,8 NLI-based Zero Shot Text Classification 9,10 and automatic rule discovery 11 , but are also grounded in deep theoretical understanding of human psychology and social emotions. It must be explained in a nutshell that we
Human interlocutors may use emotions as an important signaling device for coordinating an interac... more Human interlocutors may use emotions as an important signaling device for coordinating an interaction. In this context, predicting a significant change in a speaker’s emotion may be important for regulating the interaction. Given the nonlinear and noisy nature of human conversations and relatively short time series they produce, such a predictive model is an open challenge, both for modeling human behavior and in engineering artificial intelligence systems for predicting change. In this paper, we present simple and theoretically grounded models for predicting the direction of change in emotion during conversation. We tested our approach on textual data from several massive conversations corpora and two different cultures: Chinese (Mandarin) and American (English). The results converge in suggesting that change in emotion may be successfully predicted, even with regard to very short, nonlinear, and noisy interactions.
Revenge is a powerful motivating force reported to underlie the behavior of various solo perpetra... more Revenge is a powerful motivating force reported to underlie the behavior of various solo perpetrators, from school shooters to right wing terrorists. In this paper, we develop an automated methodology for identifying vengeful themes in textual data. Testing the model on four datasets (vengeful texts from social media, school shooters, Right Wing terrorist and Islamic terrorists), we present promising results, even when the methodology is tested on extremely imbalanced datasets. The paper not only presents a simple and powerful methodology that may be used for the screening of solo perpetrators but also validate the simple theoretical model of revenge. https://arxiv.org/abs/2205.01731
Language, like any other semiotic code, is a sign system whose forms (words, phrases, sentences, ... more Language, like any other semiotic code, is a sign system whose forms (words, phrases, sentences, etc.) are governed by the standing for principle (SFP). However, the SFP is more complex to envision when ‘composite forms’ such as novels and dialogues are involved. In this article, and following the work of Bakhtin and Gasset, we present the idea that a composite semiotic form, such as a paragraph in a novel, does not stand for a simple referential situation only, but provides us with a context for anticipation. We show how these ideas can be expressed in the state-of-the-art language models in AI (Artificial Intelligence) and the way AI may provide us with novel directions for developing old practices of interpretation.
The identification of extreme rare events is a challenge that appears in several real-world conte... more The identification of extreme rare events is a challenge that appears in several real-world contexts, from screening for solo perpetrators to the prediction of failures in industrial production. In this article, we explain the challenge and present a new methodology for addressing it, a methodology that may be considered in terms of features engineering. This methodology, which is based on Jaynes inferential approach, is tested on a dataset dealing with failures in production in the pulp-and-paper industry. The results are discussed in the context of the benefits of using the approach for features engineering in practical contexts involving measurable risks.
Progress in Biophysics and Molecular Biology, Sep 1, 2007
We commonly think of the immune system as having a memory. However, memory is always accompanied ... more We commonly think of the immune system as having a memory. However, memory is always accompanied by a complementary process of oblivion. Is there immune oblivion? In this theoretical paper, I address this question and suggest that oblivion is an integral aspect of memorization. In this context, I suggest that immune memory is an orchestration of reversible and irreversible processes of biological computation through feedback loops. Drawing on the linguistic metaphor, I inquire into the implications of this idea for a better understanding of immune memory and immune deficiency among the elderly.
Information Sciences an International Journal, 2009
Excavating the meaning of a target term is a challenge facing information sciences and various ap... more Excavating the meaning of a target term is a challenge facing information sciences and various applications of Web 3.0. In this paper, we present a novel method for excavating the meaning of a target term by analyzing metaphors in which the term is embedded. More specifically, we (1) review some of the basic theoretical difficulties associated with the dominant cognitive
The Quarterly Journal of Experimental Psychology a Human Experimental Psychology, Aug 1, 2003
Informal reasoning fallacies are arguments that, though they may seem persuasive, are not valid. ... more Informal reasoning fallacies are arguments that, though they may seem persuasive, are not valid. The psychological aspect of informal reasoning fallacies, specifically the identification of factors that influence students’ ability to identify fallacies, has not been the subject of empirical study. The aim of this study was to test the hypothesis that subjects’ ability to identify fallacious arguments is associated with the representation of the argumentative text in the cognitive system. In the first experiment, we tested the hypothesis through a recall task. In the second experiment, we tested the hypothesis through a classification task. The results of the experiments confirm the research hypothesis and point to the role of argumentative structures in argumentation tasks.
Text Interdisciplinary Journal For the Study of Discourse, Jan 18, 2001
In this article, we discuss the meaning of rationality in rhetoric through an examination of fund... more In this article, we discuss the meaning of rationality in rhetoric through an examination of fundamentalist rhetoric. Liberal critics of fundamentalism far too often dismiss their opponents and their speech as 'irrational'. In our interpretation, fundamentalist rhetoric is neither irrational nor is it ...
Context has been a core concept for understanding human behavior and the interpretation of texts.... more Context has been a core concept for understanding human behavior and the interpretation of texts. However, the concept is not well formalized in a way that can be used by intelligent systems for performing various social computing tasks, such as identifying social norm violations. In this short paper, I present a novel way through which we may think about context when analyzing conversations and identifying social norm violations. The ideas and methodology are illustrated through a case study where an American comedy movie script is analyzed. The paper concludes by pointing to our shortcomings in contextual representation and interpretation and by examining the relevance of Natural Language Inference for improving both human and machine interpretation for social computing and the discovery of social norm violations.
Data Augmentation for Modeling Human Personality, 2023
Modeling human personality is important for several AI challenges, from the engineering of artifi... more Modeling human personality is important for several AI challenges, from the engineering of artificial psychotherapists to the design of persona bots. However, the field of computational personality analysis heavily relies on labeled data, which may be expensive, difficult or impossible to get. This problem is amplified when dealing with rare personality types or disorders (e.g., the antisocial psychopathic personality disorder). In this context, we developed a text-based data augmentation approach for human personality (PEDANT). PEDANT doesn't rely on the common type of labeled data but on the generative pre-trained model (GPT) combined with domain expertise. Testing the methodology on three different datasets, provides results that support the quality of the generated data.
Mentalization describes the process through which we understand the mental states of oneself and ... more Mentalization describes the process through which we understand the mental states of oneself and others. In this paper, I present a computational semiotic model of mentalization and illustrate it through a worked-out example. The model draws on classical semiotic ideas, such as abductive inference and hypostatic abstraction, but pour them into new ideas and tools from natural language processing, machine learning and neural networks, to form a novel model of language-mediated-mentalization.
From the cell to the human brain, "reality" is always mediated. This postulate portrays living sy... more From the cell to the human brain, "reality" is always mediated. This postulate portrays living systems as "meaning making", because they must map sensed signals into signaling pathways through which representations and behavior are formed. In this paper, I present prolegomena to a future theory of meaning making, which is grounded in basic tenets of semiotics, while at the same time point to the way this prolegomenon may help us to understand meaning making stretching from the pre-linguistic realm upward. The idea of "meaning making" is grounded in the very basic fact that from the bacteria to the human being, reality is always mediated. Having a boundary with the environment and the need to operate on the environment are just two reasons explaining why living systems have developed various ways for representing certain portions of the environment and operating on these representations. In other words, meaning making in living systems is grounded in the basic fact that reality is always mediated and cannot be directly approached through "intuition" as clearly argued by Peirce (W2: 193). The representations formed by human and non-human organisms alike, are clearly not simple maps of "The" environment; The map is not the territory, the bacterium's representation of the environment cannot be confused with its environment and our own scientific models of reality, successful as they may be, involve only partial maps of reality. Therefore, the basic understanding of "meaning
The availability of historical textual corpora has led to the study of words’ volume (i.e. freque... more The availability of historical textual corpora has led to the study of words’ volume (i.e. frequency) along the historical time line, as representing the public’s focus of attention over time. However, study of the dynamics of words’ meaning is still in its infancy. In this paper, we first propose that the meaning of a word may be studied through the entropy of its embedding. Using two historical case studies, we show that this entropy measure is correlated with how much a word is used (i.e. its volume). More importantly, we show that using Tsallis entropy with a superadditive entropy index may provide a better estimation of a word’s volume than using Shannon entropy. We explain this finding as resulting from an increasing redundancy between the words that comprise the semantic field of the target word and develop a new measure of redundancy between words. Using this measure, which relies on the Tsallis version of the Kullback–Leibler divergence, we show that the evolving meaning of a word involves the dynamics of increasing redundancy between components of its semantic field. The proposed methodology may enrich the toolkit of researchers who study the dynamics of word senses in textual historical corpora.
Keywords: complex social systems, Tsallis entropy, word dynamics, historical corpora, multidisciplinary science
The brain's architecture is usually discussed as a hierarchy where information flows in a bottom-... more The brain's architecture is usually discussed as a hierarchy where information flows in a bottom-up manner. However, local and recurrent cortical connections form an important aspect of the brain's architecture and it is not quite clear what is the function of this "horizontal" form of connectivity. In this theoretical paper, we aim to provide one possible explanation to this question using the perspective of abstract algebra and more specifically the one of Category Theory. We propose that recurrent interactions can be modeled by using an abstract structure which is the Groupoid. This abstract structure may explain the role of recurrent local circuits in natural computing machines with multi layers of representations.
A final draft of a paper that has been accepted for publication in PlosOne
Abstract
The aim of t... more A final draft of a paper that has been accepted for publication in PlosOne
Abstract The aim of this study was to analyze dynamic patterns for scanning femoroacetabular impingement (FAI) radiographs in orthopedics, in order to better understand the nature of expertise in radiography. Seven orthopedics residents with at least two years of expertise and seven board-certified orthopedists participated in the study. The participants were asked to diagnose 15 anteroposterior (AP) pelvis radiographs of 15 surgical patients, diagnosed with FAI syndrome. Eye tracking data were recorded using the SMI desk-mounted tracker and were analyzed using advanced measures and methodologies, mainly recurrence quantification analysis. The expert orthopedists presented a less predictable pattern of scanning the radiographs although there was no difference between experts and non-experts in the deterministic nature of their scan path. In addition, the experts presented a higher percentage of correct areas of focus and more quickly made their first comparison between symmetric regions of the pelvis. We contribute to the understanding of experts’ process of diagnosis by showing that experts are qualitatively different from residents in their scanning patterns. The dynamic pattern of scanning that characterizes the experts was found to have a more complex and less predictable signature, meaning that experts’ scanning is simultaneously both structured (i.e. deterministic) and unpredictable.
The idea that abstract words are grounded in our sensorimotor experience is gaining support and p... more The idea that abstract words are grounded in our sensorimotor experience is gaining support and popularity, as observed in the increasing number of studies dealing with "neurosemantics." Therefore, it is important to form models that explain how to bridge the gap between basic bodily experiences and abstract language. This paper focuses on the embodiment of connotations, such as 'sweet' in 'sweet baby', where the adjective has been abstracted from its concrete and embodied sense. We summarize several findings from recent studies in neuroscience and the cognitive sciences suggesting that emotion, body, and language are three factors required for understanding the emergence of abstract words, and (1) propose a model explaining how these factors contribute to the emergence of connotations, (2) formulate a computational model instantiating our theoretical model and (3) test our model in a task involving the automatic identification of connotations. The results support our model pointing to the role of embodiment in the formation of connotations.
Uploads
Drawing on years of R&D and academic publications in top rated journals, the author provides the reader with a simple and deep understanding of context and its modeling for specific challenges, from identifying social norm violations to understanding conversations going awry and stories by great authors. The book may interest a wide variety of readers seeking to incorporate AI into their understanding of human behavior.
The book outlines the affordances of new technologies for textual analysis, which has historically employed established approaches within the humanities. Neuman, Danesi, and Vilenchik argue that these different forms of analysis are indeed complementary, demonstrating the ways in which AI-based perspectives echo similar theoretical and methodological currents in traditional approaches while also offering new directions for research. The volume showcases examples from a wide range of texts, including novels, television shows, and films to illustrate the ways in which the latest AI technologies can be used for "dialoguing" with textual characters and examining textual meaning coherence.
Illuminating the potential of AI language models to both enhance and extend research on the interpretation of texts, this book will appeal to scholars interested in cognitive approaches to the humanities in such fields as literary studies, discourse analysis, media studies, film studies, psychology, and artificial intelligence.
https://www.routledge.com/How-to-Find-a-Needle-in-a-Haystack-From-the-Insider-Threat-to-Solo-Perpetrators/Neuman/p/book/9781032229768#
Searching for a needle in a haystack is an important task in several contexts of data analysis and decision-making. Examples include identifying the insider threat within an organization, the prediction of failure in industrial production, or pinpointing the unique signature of a solo perpetrator, such as a school shooter or a lone wolf terrorist. It is a challenge different from that of identifying a rare event (e.g., a tsunami) or detecting anomalies because the "needle" is not easily distinguished from the haystack. This challenging context is imbued with particular difficulties, from the lack of sufficient data to train a machine learning model through the identification of the relevant features and up to the painful price of false alarms, which might cause us to question the relevance of machine learning solutions even if they perform well according to common performance criteria. In this book, Prof. Neuman approaches the problem of finding the needle by specifically focusing on the human factor, from solo perpetrators to insider threats. Providing for the first time a deep, critical, multidimensional, and methodological analysis of the challenge, the book offers data scientists and decision makers a deep scientific foundational approach combined with a pragmatic practical approach that may guide them in searching for a needle in a haystack.
"Conceptual Mathematics and Literature employs an interdisciplinary mathematical approach that is a uniquely insightful interpretive analysis of literary works. It offers novel ways of comprehending previously unnoticed underlying patterns of meaning. Numerous figures illustrate and reinforce the author's brilliant and illuminating mode of critical inquiry."-Frank Nuessel, Professor of Modern Languages and Linguistics, University Scholar at the University of Louisville
According to the central thesis of biosemiotics, sign processes characterise all living systems and the very nature of life, and their diverse phenomena can be best explained via the dynamics and typology of sign relations. The authors are therefore presenting a deeper view on biological evolution, intentionality of organisms, the role of communication in the living world and the nature of sign systems — all topics which are described in this volume. This has important consequences on the methodology and epistemology of biology and study of life phenomena in general, which the authors aim to help the reader better understand."
https://arxiv.org/abs/2205.01731
Drawing on years of R&D and academic publications in top rated journals, the author provides the reader with a simple and deep understanding of context and its modeling for specific challenges, from identifying social norm violations to understanding conversations going awry and stories by great authors. The book may interest a wide variety of readers seeking to incorporate AI into their understanding of human behavior.
The book outlines the affordances of new technologies for textual analysis, which has historically employed established approaches within the humanities. Neuman, Danesi, and Vilenchik argue that these different forms of analysis are indeed complementary, demonstrating the ways in which AI-based perspectives echo similar theoretical and methodological currents in traditional approaches while also offering new directions for research. The volume showcases examples from a wide range of texts, including novels, television shows, and films to illustrate the ways in which the latest AI technologies can be used for "dialoguing" with textual characters and examining textual meaning coherence.
Illuminating the potential of AI language models to both enhance and extend research on the interpretation of texts, this book will appeal to scholars interested in cognitive approaches to the humanities in such fields as literary studies, discourse analysis, media studies, film studies, psychology, and artificial intelligence.
https://www.routledge.com/How-to-Find-a-Needle-in-a-Haystack-From-the-Insider-Threat-to-Solo-Perpetrators/Neuman/p/book/9781032229768#
Searching for a needle in a haystack is an important task in several contexts of data analysis and decision-making. Examples include identifying the insider threat within an organization, the prediction of failure in industrial production, or pinpointing the unique signature of a solo perpetrator, such as a school shooter or a lone wolf terrorist. It is a challenge different from that of identifying a rare event (e.g., a tsunami) or detecting anomalies because the "needle" is not easily distinguished from the haystack. This challenging context is imbued with particular difficulties, from the lack of sufficient data to train a machine learning model through the identification of the relevant features and up to the painful price of false alarms, which might cause us to question the relevance of machine learning solutions even if they perform well according to common performance criteria. In this book, Prof. Neuman approaches the problem of finding the needle by specifically focusing on the human factor, from solo perpetrators to insider threats. Providing for the first time a deep, critical, multidimensional, and methodological analysis of the challenge, the book offers data scientists and decision makers a deep scientific foundational approach combined with a pragmatic practical approach that may guide them in searching for a needle in a haystack.
"Conceptual Mathematics and Literature employs an interdisciplinary mathematical approach that is a uniquely insightful interpretive analysis of literary works. It offers novel ways of comprehending previously unnoticed underlying patterns of meaning. Numerous figures illustrate and reinforce the author's brilliant and illuminating mode of critical inquiry."-Frank Nuessel, Professor of Modern Languages and Linguistics, University Scholar at the University of Louisville
According to the central thesis of biosemiotics, sign processes characterise all living systems and the very nature of life, and their diverse phenomena can be best explained via the dynamics and typology of sign relations. The authors are therefore presenting a deeper view on biological evolution, intentionality of organisms, the role of communication in the living world and the nature of sign systems — all topics which are described in this volume. This has important consequences on the methodology and epistemology of biology and study of life phenomena in general, which the authors aim to help the reader better understand."
https://arxiv.org/abs/2205.01731
Keywords: complex social systems, Tsallis entropy, word dynamics, historical corpora, multidisciplinary science
* This paper has been submitted for publication
Abstract
The aim of this study was to analyze dynamic patterns for scanning femoroacetabular impingement (FAI) radiographs in orthopedics, in order to better understand the nature of expertise in radiography. Seven orthopedics residents with at least two years of expertise and seven board-certified orthopedists participated in the study. The participants were asked to diagnose 15 anteroposterior (AP) pelvis radiographs of 15 surgical patients, diagnosed with FAI syndrome. Eye tracking data were recorded using the SMI desk-mounted tracker and were analyzed using advanced measures and methodologies, mainly recurrence quantification analysis.
The expert orthopedists presented a less predictable pattern of scanning the radiographs although there was no difference between experts and non-experts in the deterministic nature of their scan path. In addition, the experts presented a higher percentage of correct areas of focus and more quickly made their first comparison between symmetric regions of the pelvis. We contribute to the understanding of experts’ process of diagnosis by showing that experts are qualitatively different from residents in their scanning patterns. The dynamic pattern of scanning that characterizes the experts was found to have a more complex and less predictable signature, meaning that experts’ scanning is simultaneously both structured (i.e. deterministic) and unpredictable.