Search

There exists a growing discourse around the domination of Big Tech on the landscape of artificial intelligence (AI) research, yet our comprehension of this phenomenon remains cursory. This paper aims to broaden and deepen our understanding of Big Tech's reach and power within AI research. It highlights the dominance not merely in terms of sheer publication volume but rather in the propagation of new ideas or memes. Current studies often oversimplify the concept of influence to the share of affiliations in academic papers, typically sourced from limited databases such as arXiv or specific academic conferences. The main goal of this paper is to unravel the specific nuances of such influence, determining which AI ideas are predominantly driven by Big Tech entities. By employing network and memetic analysis on AI-oriented paper abstracts and their citation network, we are able to grasp a deeper insight into this phenomenon. By utilizing two databases: OpenAlex and S2ORC, we are able to perform such analysis on a much bigger scale than previous attempts. Our findings suggest that while Big Tech-affiliated papers are disproportionately more cited in some areas, the most cited papers are those affiliated with both Big Tech and Academia. Focusing on the most contagious memes, their attribution to specific affiliation groups (Big Tech, Academia, mixed affiliation) seems equally distributed between those three groups. This suggests that the notion of Big Tech domination over AI research is oversimplified in the discourse.

In this paper, we examine the interplay between the lobby size $q$ in the $q$-neighbor Ising model of opinion formation [Phys. Rev. E {\bf 92}, 052105] and the level of overlap $v$ of two fully connected graphs. Results suggest that for each lobby size $q \ge 3$, a specific level of overlap $v^*$ exists, which destroys initially polarized clusters of opinions. By performing Monte-Carlo simulations, backed by an analytical approach, we show that the dependence of the $v^*$ on the lobby size $q$ is far from trivial in the absence of temperature, showing consecutive maximum and minimum, that additionally depends on the parity of $q$. The temperature is, in general, a destructive factor; its increase leads to the collapse of polarized clusters for smaller values of $v$ and additionally brings a substantial decrease in the level of polarization. However, we show that this behavior is counter-intuitively inverted for specific lobby sizes and temperature ranges., Comment: 12 pages, 6 figures

This paper introduces HADES, a novel tool for automatic comparative documents with similar structures. HADES is designed to streamline the work of professionals dealing with large volumes of documents, such as policy documents, legal acts, and scientific papers. The tool employs a multi-step pipeline that begins with processing PDF documents using topic modeling, summarization, and analysis of the most important words for each topic. The process concludes with an interactive web app with visualizations that facilitate the comparison of the documents. HADES has the potential to significantly improve the productivity of professionals dealing with high volumes of documents, reducing the time and effort required to complete tasks related to comparative document analysis. Our package is publically available on GitHub.

The number of standardized policy documents regarding climate policy and their publication frequency is significantly increasing. The documents are long and tedious for manual analysis, especially for policy experts, lawmakers, and citizens who lack access or domain expertise to utilize data analytics tools. Potential consequences of such a situation include reduced citizen governance and involvement in climate policies and an overall surge in analytics costs, rendering less accessibility for the public. In this work, we use a Latent Dirichlet Allocation-based pipeline for the automatic summarization and analysis of 10-years of national energy and climate plans (NECPs) for the period from 2021 to 2030, established by 27 Member States of the European Union. We focus on analyzing policy framing, the language used to describe specific issues, to detect essential nuances in the way governments frame their climate policies and achieve climate goals. The methods leverage topic modeling and clustering for the comparative analysis of policy documents across different countries. It allows for easier integration in potential user-friendly applications for the development of theories and processes of climate policy. This would further lead to better citizen governance and engagement over climate policies and public policy research., Comment: Accepted for Tackling Climate Change with Machine Learning: workshop at NeurIPS 2022

The growing number of AI applications, also for high-stake decisions, increases the interest in Explainable and Interpretable Machine Learning (XI-ML). This trend can be seen both in the increasing number of regulations and strategies for developing trustworthy AI and the growing number of scientific papers dedicated to this topic. To ensure the sustainable development of AI, it is essential to understand the dynamics of the impact of regulation on research papers as well as the impact of scientific discourse on AI-related policies. This paper introduces a novel framework for joint analysis of AI-related policy documents and eXplainable Artificial Intelligence (XAI) research papers. The collected documents are enriched with metadata and interconnections, using various NLP methods combined with a methodology inspired by Institutional Grammar. Based on the information extracted from collected documents, we showcase a series of analyses that help understand interactions, similarities, and differences between documents at different stages of institutionalization. To the best of our knowledge, this is the first work to use automatic language analysis tools to understand the dynamics between XI-ML methods and regulations. We believe that such a system contributes to better cooperation between XAI researchers and AI policymakers.

Echo chambers and polarisation dynamics are as of late a very prominent topic in scientific communities around the world. As these phenomena directly affect our lives and seemingly more and more as our societies and communication channels evolve it becomes ever so important for us to understand the intricacies of opinion dynamics in the modern era. Here we extend an existing echo chambers model with activity driven agents onto a bi-layer topology and study the dynamics of the polarised state as a function of interlayer couplings. Different cases of such couplings are presented - unidirectional coupling that can be reduced to a mono-layer facing an external bias, symmetric and non-symmetric couplings. We have assumed that initial conditions impose system polarisation and agent opinions are different for both layers. Such a pre-conditioned polarised state can sustain without explicit homophilic interactions provided the coupling strength between agents belonging to different layers is weak enough. For a strong unidirectional or attractive coupling between two layers a discontinuous transition to a radicalised state takes place when mean opinions in both layers are the same. When coupling constants between the layers are of different signs the system exhibits sustained or decaying oscillations. Transitions between these states are analysed using a mean field approximation and classified in the framework of bifurcation theory.

Finding hidden layers in complex networks is an important and a non-trivial problem in modern science. We explore the framework of quantum graphs to determine whether concealed parts of a multi-layer system exist and if so then what is their extent, i.e., how many unknown layers there are. Assuming that all information available is the time evolution of a wave propagation on a single layer of a network it is indeed possible to uncover that which is hidden by merely observing the dynamics. We present evidence on both synthetic and real-world networks that the frequency spectrum of the wave dynamics can express distinct features in the form of additional frequency peaks. These peaks exhibit dependence on the number of layers taking part in the propagation and thus allowing for the extraction of said number. We show that in fact, with sufficient observation time, one can fully reconstruct the row-normalised adjacency matrix spectrum. We compare our propositions to a machine learning approach using a modified, for the purposes of multi-layer systems, wave packet signature method., Comment: 20 pages, 13 figures

Urban mobility needs alternative sustainable travel modes to keep our pandemic cities in motion. Ride-pooling, where a single vehicle is shared by more than one traveller, is not only appealing for mobility platforms and their travellers, but also for promoting the sustainability of urban mobility systems. Yet, the potential of ride-pooling rides to serve as a safe and effective alternative given the personal and public health risks considerations associated with the COVID-19 pandemic is hitherto unknown. To answer this, we combine epidemiological and behavioural shareability models to examine spreading among ride-pooling travellers, with an application for Amsterdam. Findings are at first sight devastating, with only few initially infected travellers needed to spread the virus to hundreds of ride-pooling users. Without intervention, ride-pooling system may substantially contribute to virus spreading. Notwithstanding, we identify an effective control measure allowing to halt the spreading before the outbreaks (at 50 instead of 800 infections) without sacrificing the efficiency achieved by pooling. Fixed matches among co-travellers disconnect the otherwise dense contact network, encapsulating the virus in small communities and preventing the outbreaks.

We analyze a nonlinear $q$-voter model with stochastic noise, interpreted in the social context as independence, on a duplex network. The size of the lobby $q$ (i.e., the pressure group) is a crucial parameter that changes the behavior of the system. The $q$-voter model has been applied on multiplex networks in a previous work [Phys. Rev E. 92. 052812. (2015)], and it has been shown that the character of the phase transition depends on the number of levels in the multiplex network as well as the value of $q$. Here we study phase transition character in the case when on each level of the network the lobby size is different, resulting in two parameters $q_1$ and $q_2$. We find evidence of successive phase transitions when a continuous phase transition is followed by a discontinuous one or two consecutive discontinuous phases appear, depending on the parameter. When analyzing this system, we even encounter mixed-order (or hybrid) phase transition. We perform simulations and obtain supporting analytical solutions on a simple multiplex case - a duplex clique, which consists of two fully overlapped complete graphs (cliques)., Comment: 13 pages, 10 figures

Using a set of over 70.000 records from PLOS One journal consisting of 37 lexical, sentiment and bibliographic variables we perform analysis backed with machine learning methods to predict the class of popularity of scientific papers defined by the number of times they have been viewed. Our study shows correlations among the features and recovers a threshold for the number of views that results in the best prediction results in terms of Matthew's correlation coefficient. Moreover, by creating a variable importance plot for random forest classifier, we are able to reduce the number of features while keeping similar predictability and determine crucial factors responsible for the popularity., Comment: 13 pages, 6 figures

The Hamiltonian Mean-Field (HMF) model belongs to a broad class of statistical physics models with non-additive Hamiltonians that reveal many non-trivial properties, such as non-equivalence of statistical ensembles, ergodicity breaking, and negative specific heat. With this paper, we add to this set another intriguing feature, which is that of super-diffusive equilibrium dynamics. Using molecular dynamics techniques, we compare the diffusive properties of the HMF model in the quasi-stationary metastable state (QSS) and in the Boltzmann-Gibbs (BG) regime. In contrast to the current state of knowledge, we show that L\'evy walks underlying super-diffusion in QSS do not disappear when the system settles in the thermodynamic equilibrium. We demonstrate that it is extremely difficult to distinguish QSS from the BG regime, by only examining the statistics of L\'evy walks in HMF particle trajectories. We construct a simple stochastic model based on the truncated L\'{e}vy walks with rests that quantitatively resembles diffusion behavior observed in both stages of the HMF dynamics.

We show that activity of online news outlets follows a temporal fluctuation scaling law and we recover this feature using an independent cascade model augmented with a varying hype parameter representing a viral potential of an original article. We use the Event Registry platform to track activity of over 10,000 news outlets in 11 different topics in the course of the year 2016. Analyzing over 22,000,000 articles, we found that fluctuation scaling exponents $\alpha$ depend on time window size $\Delta$ in a characteristic way for all the considered topics -- news outlets activities are partially synchronized for $\Delta>15\mathrm{min}$ with a cross-over for $\Delta=1\mathrm{day}$. The proposed model was run on several synthetic network models as well as on a network extracted from the real data. Our approach discards timestamps as not fully reliable observables and focuses on co-occurrences of publishers in cascades of similarly phrased news items. We make use of the Event Registry news clustering feature to find correlations between content published by news outlets in order to uncover common information propagation paths in published articles and to estimate weights of edges in the independent cascade model. While the independent cascade model follows the fluctuation scaling law with a trivial exponent $\alpha=0.5$, we argue that besides the topology of the underlying cooperation network a temporal clustering of articles with similar hypes is necessary to qualitatively reproduce the fluctuation scaling observed in the data.

We analyze a modified kinetic Ising model, so called $q$-neighbor Ising model, with Metropolis dynamics, [Phys. Rev. E 92, 052105], on a duplex clique and a partially duplex clique. In the $q$-neighbor Ising model each spin interacts only with $q$ spins randomly chosen from its whole neighborhood. In the case of a duplex clique the change of a spin is allowed only if both levels simultaneously induce this change. Due to the mean-field like nature of the model we are able to derive the analytic form of transition probabilities and solve the corresponding master equation. The existence of the second level changes dramatically the character of the phase transition. In the case of the monoplex clique, the $q$-neighbor Ising model exhibits continuous phase transition for $q=3$, discontinuous phase transition for $q \ge 4$ and for $q=1$ and $q=2$ the phase transition is not observed. On the other hand, in the case of the duplex clique continuous phase transitions are observed for all values of $q$, even for $q=1$ and $q=2$. Subsequently we introduce a partially duplex clique, parametrized by $r \in [0,1]$, which allows us to tune the network from monoplex ($r=0$) to duplex ($r=1$). Such a generalized topology, in which a fraction $r$ of all nodes appear on both levels, allows to obtain the critical value of $r=r^*(q)$ at which a tricriticality (switch from continuous to discontinuous phase transition) appears., Comment: 13 pages, 9 figures, 3 appendices

We investigate how textual properties of scientific papers relate to the number of citations they receive. Our main finding is that correlations are non-linear and affect differently most-cited and typical papers. For instance, we find that in most journals short titles correlate positively with citations only for the most cited papers, for typical papers the correlation is in most cases negative. Our analysis of 6 different factors, calculated both at the title and abstract level of 4.3 million papers in over 1500 journals, reveals the number of authors, and the length and complexity of the abstract, as having the strongest (positive) influence on the number of citations., Comment: 9 pages, 3 figures, 3 tables

High frequency psychophysiological data create a challenge for quantitative modeling based on Big Data tools since they reflect the complexity of processes taking place in human body and its responses to external events. Here we present studies of fluctuations in facial electromyography (fEMG) and electrodermal activity (EDA) massive time series and changes of such signals in the course of emotional stimulation. Zygomaticus major (ZYG, "smiling" muscle) activity, corrugator supercilii (COR, "frowning"bmuscle) activity, and phasic skin conductance (PHSC, sweating) levels of 65 participants were recorded during experiments that involved exposure to emotional stimuli (i.e., IAPS images, reading and writing messages on an artificial online discussion board). Temporal Taylor's fluctuations scaling were found when signals for various participants and during various types of emotional events were compared. Values of scaling exponents were close to 1, suggesting an external origin of system dynamics and/or strong interactions between system's basic elements (e.g., muscle fibres). Our statistical analysis shows that the scaling exponents enable identification of high valence and arousal levels in ZYG and COR signals.

We demonstrate the power of data mining techniques for the analysis of collective social dynamics within British Tweets during the Olympic Games 2012. The classification accuracy of online activities related to the successes of British athletes significantly improved when emotional components of tweets were taken into account, but employing emotional variables for activity prediction decreased the classifiers' quality. The approach could be easily adopted for any prediction or classification study with a set of problem-specific variables., Comment: 16 pages, 9 figures, 2 tables and 1 appendix

We solve the growing asymmetric Ising model [Phys. Rev. E 89, 012105 (2014)] in the topologies of deterministic and stochastic (random) scale-free trees predicting its non-monotonous behavior for external fields smaller than the coupling constant $J$. In both cases we indicate that the crossover temperature corresponding to maximal magnetization decays approximately as $(\ln \ln N)^{-1}$, where $N$ is the number of nodes in the tree., Comment: 6 pages, 6 figures

We perform the analysis of scientific collaboration at the level of universities. The scope of this study is to answer two fundamental questions: (i) can one indicate a category (i.e., a scientific discipline) that has the greatest impact on the rank of the university and (ii) do the best universities collaborate with the best ones only? Using two university ranking lists (ARWU and QS) as well as data from the Science Citation Index we show how the number of publications in certain categories correlates with the university rank. Moreover, using complex networks analysis, we give hints that the scientific collaboration is highly embedded in the physical space and the number of common papers decays with the distance between them. We also show the strength of the ties between universities is proportional to product of their total number of publications., Comment: 17 pages, 12 figures, 3 tables

We introduce a growing one-dimensional quenched spin model that bases on asymmetrical one-side Ising interactions in the presence of external field. Numerical simulations and analytical calculations based on Markov chain theory show that when the external field is smaller than the exchange coupling constant $J$ there is a non-monotonous dependence of the mean magnetization on the temperature in a finite system. The crossover temperature $T_c$ corresponding to the maximal magnetization decays with system size, approximately as the inverse of the W Lambert function. The observed phenomenon can be understood as an interplay between the thermal fluctuations and the presence of the first cluster determined by initial conditions. The effect exists also when spins are not quenched but fully thermalized after the attachment to the chain. We conceive the model is suitable for a qualitative description of online emotional discussions arranged in a chronological order, where a spin in every node conveys emotional valence of a subsequent post., Comment: 8 pages, 9 figures

We performed statistical analysis on data from the Digg.com website, which enables its users to express their opinion on news stories by taking part in forum-like discussions as well as directly evaluate previous posts and stories by assigning so called "diggs". Owing to fact that the content of each post has been annotated with its emotional value, apart from the strictly structural properties, the study also includes an analysis of the average emotional response of the posts commenting the main story. While analysing correlations at the story level, an interesting relationship between the number of diggs and the number of comments received by a story was found. The correlation between the two quantities is high for data where small threads dominate and consistently decreases for longer threads. However, while the correlation of the number of diggs and the average emotional response tends to grow for longer threads, correlations between numbers of comments and the average emotional response are almost zero. We also show that the initial set of comments given to a story has a substantial impact on the further "life" of the discussion: high negative average emotions in the first 10 comments lead to longer threads while the opposite situation results in shorter discussions. We also suggest presence of two different mechanisms governing the evolution of the discussion and, consequently, its length., Comment: 26 pages, 16 figures, 6 tables

We analyze emotionally annotated massive data from IRC (Internet Relay Chat) and model the dialogues between its participants by assuming that the driving force for the discussion is the entropy growth of emotional probability distribution. This process is claimed to be correlated to the emergence of the power-law distribution of the discussion lengths observed in the dialogues. We perform numerical simulations based on the noticed phenomenon obtaining a good agreement with the real data. Finally, we propose a method to artificially prolong the duration of the discussion that relies on the entropy of emotional probability distribution., Comment: 9 pages, 2 tables

We perform a statistical analysis of emotionally annotated comments in two large online datasets, examining chains of consecutive posts in the discussions. Using comparisons with randomised data we show that there is a high level of correlation for the emotional content of messages., Comment: REVTEX format, 5 pages, 6 figures, 2 tables, accepted to Acta Physica Polonica A

E-communities, social groups interacting online, have recently become an object of interdisciplinary research. As with face-to-face meetings, Internet exchanges may not only include factual information but also emotional information - how participants feel about the subject discussed or other group members. Emotions are known to be important in affecting interaction partners in offline communication in many ways. Could emotions in Internet exchanges affect others and systematically influence quantitative and qualitative aspects of the trajectory of e-communities? The development of automatic sentiment analysis has made large scale emotion detection and analysis possible using text messages collected from the web. It is not clear if emotions in e-communities primarily derive from individual group members' personalities or if they result from intra-group interactions, and whether they influence group activities. We show the collective character of affective phenomena on a large scale as observed in 4 million posts downloaded from Blogs, Digg and BBC forums. To test whether the emotions of a community member may influence the emotions of others, posts were grouped into clusters of messages with similar emotional valences. The frequency of long clusters was much higher than it would be if emotions occurred at random. Distributions for cluster lengths can be explained by preferential processes because conditional probabilities for consecutive messages grow as a power law with cluster length. For BBC forum threads, average discussion lengths were higher for larger values of absolute average emotional valence in the first ten comments and the average amount of emotion in messages fell during discussions. Our results prove that collective emotional states can be created and modulated via Internet communication and that emotional expressiveness is the fuel that sustains some e-communities., Comment: 23 pages including Supporting Information, accepted to PLoS ONE

We present an empirical study of user activity in online BBC discussion forums, measured by the number of posts written by individual debaters and the average sentiment of these posts. Nearly 2.5 million posts from over 18 thousand users were investigated. Scale free distributions were observed for activity in individual discussion threads as well as for overall activity. The number of unique users in a thread normalized by the thread length decays with thread length, suggesting that thread life is sustained by mutual discussions rather than by independent comments. Automatic sentiment analysis shows that most posts contain negative emotions and the most active users in individual threads express predominantly negative sentiments. It follows that the average emotion of longer threads is more negative and that threads can be sustained by negative comments. An agent based computer simulation model has been used to reproduce several essential characteristics of the analyzed system. The model stresses the role of discussions between users, especially emotionally laden quarrels between supporters of opposite opinions, and represents many observed statistics of the forum., Comment: 29 pages, 6 figures

We extend a model of community isolation in the d-dimensional lattice onto the case with an imposed imbalance between birth rates of competing communities. We give analytical and numerical evidences that in the asymmetric two-specie model there exists a well defined value of the asymmetry parameter when the emergence of the isolated (blocked) subgroups is the fastest, i.e. the characteristic time tc is minimal. This critical value of the parameter depends only on the lattice dimensionality and is independent from the system size. Similar phenomenon was observed in the multi-specie case with a geometric distribution of the birth rates. We also show that blocked subgroups in the multi-specie case are absent or very rare when either there is a strictly dominant specie that outnumbers the others or when there is a large diversity of species. The number of blocked species of different kinds decreases with the dimension of the multi-specie system., Comment: 6 pages, 4 figures

We extend the previously observed scaling equation connecting the internode distances and nodes' degrees onto the case of weighted networks. We show that the scaling takes a similar form in the empirical data obtained from networks characterized by different relations between node's strength and its degree. In the case of explicit equation for s(k) (e.g. linear or scale-free), the new coefficients of scaling equation can be easily obtained. We support our analysis with numerical simulations for Erdos-Renyi random graphs with different weight distributions., Comment: 9 pages, 4 figures, submitted to International Journal of Modern Physics C

We introduce a simple model of a growing system with $m$ competing communities. The model corresponds to the phenomenon of defeats suffered by social groups living in isolation. A nonequilibrium phase transition is observed when at critical time $t_c$ the first isolated cluster occurs. In the one-dimensional system the volume of the new phase, i.e. the number of the isolated individuals, increases with time as $Z \sim t^3$. For a large number of possible communities the critical density of filled space equals to $\rho_c = (m/N)^{1/3}$ where $N$ is the system size. A similar transition is observed for Erd\H{o}s-R\'{e}nyi random graphs and Barab\'{a}si-Albert scale-free networks. Analytic results are in agreement with numerical simulations., Comment: 5 pages, 4 figures

In this study we consider relations between companies in Poland taking into account common branches they belong to. It is clear that companies belonging to the same branch compete for similar customers, so the market induces correlations between them. On the other hand two branches can be related by companies acting in both of them. To remove weak, accidental links we shall use a concept of threshold filtering for weighted networks where a link weight corresponds to a number of existing connections (common companies or branches) between a pair of nodes., Comment: 13 pages, 10 figures and 4 tables

We show that discretization of internode distribution in complex networks affects internode distances l_ij calculated as a function of degrees (k_i k_j) and an average path length as function of network size N. For dense networks there are log-periodic oscillations of above quantities. We present real-world examples of such a behavior as well as we derive analytical expressions and compare them to numerical simulations. We consider a simple case of network optimization problem, arguing that discrete effects can lead to a nontrivial solution., Comment: 5 pages, 5 figures, REVTEX

Public transport systems in 22 Polish cities have been analyzed. Sizes of these networks range from N=152 to N=2881. Depending on the assumed definition of network topology the degree distribution can follow a power law or can be described by an exponential function. Distributions of paths in all considered networks are given by asymmetric, unimodal functions. Clustering, assortativity and betweenness are studied. All considered networks exhibit small world behavior and are hierarchically organized. A transition between dissortative small networks N<=500 and assortative large networks N>=500 is observed., Comment: 11 pages, 17 figures, 2 tables, REVTEX4 format

We have examined a topology of 21 public transport networks in Poland. Our data exhibit several universal features in considered systems when they are analyzed from the point of view of evolving networks. Depending on the assumed definition of a network topology the degree distribution can follow a power law p(k) ~ k^(-\gamma) or can be described by an exponential function p(k) \~exp(-\alpha k). In the first case one observes that mean distances between two nodes are a linear function of logarithms of their degrees product., Comment: 8 pages, 5 figures, 1 table, to be published in Acta Physica Polonica

Universal scaling of distances between vertices of Erdos-Renyi random graphs, scale-free Barabasi-Albert models, science collaboration networks, biological networks, Internet Autonomous Systems and public transport networks are observed. A mean distance between two nodes of degrees k_i and k_j equals to =A-B log(k_i k_j). The scaling is valid over several decades. A simple theory for the appearance of this scaling is presented. Parameters A and B depend on the mean value of a node degree _nn calculated for the nearest neighbors and on network clustering coefficients., Comment: 4 pages, 3 figures, 1 table

The intrinsic length scales of a reaction diffusion system (Gierer-Meinhardt model)is varied by quasi-statically changing the diffusion constant of the activator and a transition from rolls to hexagon is detected. The transition is hysteretic or first order like. From stability analysis, we also analytically show the possibility of such transitions., Comment: One .tex file and four .ps files

Higher order clustering coefficients $C(x)$ are introduced for random networks. The coefficients express probabilities that the shortest distance between any two nearest neighbours of a certain vertex $i$ equals $x$, when one neglects all paths crossing the node $i$. Using $C(x)$ we found that in the Barab\'{a}si-Albert (BA) model the average shortest path length in a node's neighbourhood is smaller than the equivalent quantity of the whole network and the remainder depends only on the network parameter $m$. Our results show that small values of the standard clustering coefficient in large BA networks are due to random character of the nearest neighbourhood of vertices in such networks., Comment: 10 pages, 4 figures