COMPLEXIS 2024 Abstracts

Area 1 - Complexity in Biology and Biomedical Engineering
Full Papers Short Papers
Area 2 - Complexity in Social Sciences
Full Papers Short Papers
Area 3 - Complexity in AI/Edge/Fog/High-Performance Computing
Short Papers
Area 4 - Complexity in Informatics and Networking
Short Papers
Area 5 - Complexity in Risk and Predictive Modeling
Full Papers Short Papers

Area 1 - Complexity in Biology and Biomedical Engineering

Full Papers
Paper Nr: 26
Title:

Evaluating the Multifactorial Effects on SARS-CoV-2 Spread in Tokyo Metropolitan Area with an Agent-Based Model
Authors:

Jianing Chu and Yu Chen
Abstract: The eighth wave of Coronavirus infection in Tokyo hit high records in December 2022. This paper aims to build a Tokyo-based down-scaled simulation environment to explain the eight epidemic trends using agent-based modelling and extended SEIR denotation. Four key factors are examined in this research, that are: 1. Vaccination, 2. Virus mutation, 3. Government policy and 4. PCR test. Our investigation uncovers that the reported cases during the eight epidemic waves represent merely a fraction of the true extent of infections. Additionally, our study innovates by simulating the decline of antibodies at the individual level. Our study also innovates in combining agent-based modelling and extended SEIR modelling to simulate eight continuous epidemic waves in Tokyo, considering circumstances like Olympics, state of emergency declaration, traveling policies etc. Upon analyzing the simulated outcomes, we observe a correlation between the onset of new epidemic waves and the decrease in the population possessing antibodies. Our simulation further indicates the necessity for aligning the level of PCR testing with the available medical resources. Finally, by comparing the simulation results with actual data for the eighth wave, we forewarned of a potential resurgence in the epidemic during May and June 2023.
Download
Short Papers
Paper Nr: 19
Title:

On the Robustness of Correlation Network Models in Predicting the Safety of Bridges
Authors:

Prasad Chetti and Hesham Ali
Abstract: The problem of assessing the safety of bridges and predicting potential unacceptable deterioration levels remains one of the major problems in civil engineering. This work provides a comprehensive evaluation of the Correlation Network Model (CNM) in safety assessment and the prediction of potential safety hazards of bridges. The study applies a population analysis approach to compare individual or cluster performance against a larger set of peers. The CNM outcomes were validated using a linear regression model and a robustness analysis, resulting in a high level of consistency in identifying bridge clusters with different deterioration rates, and thereby identifying clusters with high- risk and low risk bridges. This process allows for the detection of significant parameters affecting bridge deterioration. The findings affirm the CNM’s capability in capturing complex relationships between input parameters and bridge deck conditions, which exceeds the capabilities of simple linear regression models. Furthermore, the CNM’s robustness, under various conditions and assumptions, is confirmed. The study demonstrates the potential of CNM as an effective tool for strategic planning and for efficient resource allocation, enabling focused maintenance and repair interventions on bridge infrastructures that could potentially extend their service life.
Download
Paper Nr: 18
Title:

Navigating Social Networks: A Hypergraph Approach to Influence Optimization
Authors:

Murali K. Enduri
Abstract: In this study, we introduce a novel approach to influence optimization in social networks by leveraging the mathematical framework of hypergraphs. Traditional centrality measures often fall short in capturing the multi-dimensional nature of influence. To address this gap, we propose the Spreading Influence (SI) model, a sophisticated tool designed to quantify the propagation potential of nodes more accurately within hyper-graphs. Our research embarked on a comparative analysis using the Susceptible-Infected-Recovered (SIR) model across four distinct scenarios—where the top 5, 10, 15, and 20 nodes were initially infected—in four diverse datasets: Amazon, DBLP, Email-Enron, and Cora. The SI model’s performance was benchmarked against established centrality measures: Hyperdegree Centrality (HDC), Closeness Centrality (CC), Betweenness Centrality (BC), and Hyperedge Degree Centrality (HEDC). The findings underscored the SI model’s consistently superior performance in predicting influence spread. In scenarios involving the top 10 nodes, the model exhibited up to 3.18% increased influence spread over HDC, 2 .14% over CC, 1.04% over BC, and 1.69% over HEDC. This indicates a substantial improvement in identifying key influencers within networks.
Download

Area 2 - Complexity in Social Sciences

Full Papers
Paper Nr: 12
Title:

The Robustness of a Twisted Prisoner’s Dilemma for Incorporating Memory and Unlikeliness of Occurrence
Authors:

Akihiro Takahara and Tomoko Sakiyama
Abstract: In classical game theory, because players having Defector (D) strategy tend to survive, many studies have been conducted to determine the survival of players with Cooperator (C) strategy. Recently, we have tackled the problem of the evolution of cooperators by proposing a new model called the twisted prisoner’s dilemma (TPD) model. In the proposed model, each player is given a memory length. In situations where neighbors had the same strategy as a player and a higher score than that of the player, the player updated their strategy by ignoring the classical SPD update rule. This new strategy was difficult to choose before the update. Consequently, cooperators could survive even if their memory length was small. In this study, by focusing on the system sizes, performance of the TPD model was determined. Similar results were obtained for various system sizes, except when the system size was extremely small.
Download
Paper Nr: 25
Title:

FaRS: A High-Performance Automorphism-Aware Algorithm for Graph Similarity Matching
Authors:

Fan Wang, Weiren Yu, Hai H. Wang and Victor Chang
Abstract: Role-based similarity search, predicated on the topological structure of graphs, is a highly effective and widely applicable technique for various real-world information extraction applications. Although the prominent rolebased similarity algorithm, RoleSim, successfully provides the automorphic (role) equivalence of similarity between pairs of nodes, it does not effectively differentiate nodes that exhibit exact automorphic equivalence but differ in terms of structural equivalence within a given graph. This limitation arises from disregarding most adjacency similarity information between pairs of nodes during the RoleSim computation. To address this research gap, we propose a novel single-source role similarity search algorithm, named FaRS, which employs the top Γ maximum similarity matching technique to capture more information from the classes of neighboring nodes, ensuring both automorphic equivalence and structural equivalence of role similarity. Furthermore, we establish the convergence of FaRS and demonstrate its adherence to various axioms, including uniqueness, symmetry, boundedness, and triangular inequality. Additionally, we introduce the Opt FaRS algorithm, which optimizes the computation of FaRS through two acceleration components: path extraction tracking and precomputation (P-speedup and Out-speedup approach). Experimental results on real datasets demonstrate that FaRS and Opt FaRS outperform baseline algorithms in terms of both accuracy and efficiency.
Download
Short Papers
Paper Nr: 21
Title:

Assessing the Impact of Data Governance on Decision Making in Saudi Arabia
Authors:

Bashayer A. Alotaibi, Zahyah H. Alharbi and Tahani Alqurashi
Abstract: There is currently a huge amount of data stored by Saudi Arabian organizations that requires work to make it useful. This has led to the concept of ‘data governance’ as a means of organizing data and managing its use in organizations. This research evaluates how decision making has been influenced by data governance in Saudi Arabia. Twelve interviews were conducted on two aspects of data management, data governance and data analytics, to explore how each approach affects decision making. Both interviewee groups indicated that these approaches had multiple direct and indirect effects on decision making. The interviewees mostly agreed that data governance increased confidence and trust in data and improved its quality. They viewed data governance as being likely to develop a more consistent business terminology and a set of rules and responsibilities for managing data, and that decision making would be made timelier. Despite the potential benefits of data governance for decision making, the lack of awareness about its potential makes it difficult for many Saudi Arabian organizations to benefit from its use. This study provides valuable insights for businesses considering the implementation of data governance practices to optimize their decision-making process.
Download
Paper Nr: 23
Title:

Analytical Study on Typeface Visual Identification
Authors:

Xavier Molinero, Josep Freixas and Montserrat Tàpias
Abstract: In this study, our objective is to explore methodologies for the identification of diverse typefaces. Utilizing the gathered data, we conducted a thorough analysis of the outcomes, distinguishing between successes and failures for both uppercase and lowercase letters within each typeface. The analytical framework is anchored in three distinct recognition measures. The initial measure draws upon the relative frequency of accurate responses, providing insights into the overall performance of each typeface. The second measure is constructed around the F-score derived from confusion matrices, offering a comprehensive evaluation of recognition precision and recall. Lastly, the third measure is formulated on the well-established Shapley-Shubik index, extensively scrutinized and endorsed within the realm of classical game theory. This multifaceted approach allows us to comprehensively assess the distinct aspects of typeface recognition, contributing to a nuanced understanding of their effectiveness and characteristics.
Download

Area 3 - Complexity in AI/Edge/Fog/High-Performance Computing

Short Papers
Paper Nr: 28
Title:

Generative AI for Productivity in Industry and Education
Authors:

Ferenc Héjja, Tamás Bartók, Roy Dakroub and Gergely Kocsis
Abstract: Generative AI tools are the cutting edge solutions of complex AI related problems. While investigating state-of-the-art results related to the effect of GenAI in the literature, one can note that the trends most likely lead to the expectation of a positive effect on the middle and long run. Based on these findings we define 4 productivity gain related hypotheses that we study using two types of methodologies. Namely we perform a survey research related to university-industry collaboration and quantitative studies mainly based on industrial productivity metrics. We have partnered with a major IT services provider - EPAM Systems - to be able to track, validate and analyze the key productivity metrics of software development projects, with and without using GenAI tools. This evaluation is being performed on various stages of the Software Development Lifecycle (SDLC) and on several project roles. Our goal is to measure the productivity increase provided by GenAI tools. Although this research has just started recently, considering that the area has extremely high attention we present some initial findings.
Download

Area 4 - Complexity in Informatics and Networking

Short Papers
Paper Nr: 13
Title:

Algebraic Structure of Recursively Constructed References and Its Application to Knowledge Base
Authors:

Susumu Yamasaki and Mariko Sasakura
Abstract: From management views on complex website page structures, we formulate an algebraic structure of recursively constructed page references as presenting situations of them to the website with 3-valued domain. Algebraic structure of references, abstracted from website page references, is here expressed as a finite or countably infinite set of rules, where each rule is defined, by representing the recursive relations among web page references. The situations of a reference with request to the website can be denoted as the acquisitive positive, rejective negative and suspended negative, respectively. With respect to algebraic structure, a fixed point of the mapping associated with the rule set may be a model denoting consistent evaluations to assign the situations of references to 3-valued domain. Model theory for representation of consistent evaluations of references and the rule set (constructed with references) is newly settled if a fixed point consistently exists. A retrieval derivation to detect acquisitive positives and rejective negatives can be presented, to be sound with respect to the model, based on the inference by negation as failure, which is related to the suspended negative. As multiple knowledge base formed by a tuple of rule sets, this paper next presents algebraic structure of a distributed knowledge system constrained by a state, and sequential applications of such systems, containing state transitions. Model theory can be defined with fixed point of the mapping associated with the distributed knowledge system, although the fixed point may not be always applied to modeling. If consistent fixed point modeling is available, we may have a model of the distributed knowledge system, constrained by a state. Then the application of such a distributed knowledge system may be considered as causing state transitions, following modeling and designed state transitions.
Download
Paper Nr: 15
Title:

Painter Profile Clustering Using NLP Features
Authors:

N. Y. Ilba, U. M. Yıldırım and Doruk Sen
Abstract: This study introduces a practice for clustering painter profiles using features obtained from natural language processing (NLP) techniques. The investigation of similarities among painters plays an essential function in art history. While most existing research generally focuses on the visual comparison of the artists’ work, more studies should examine the textual content available for artists. As the volume of online textual information grows, the frequency of discussions about artists and their creations has gained importance, underscoring the connection between social visibility through digital discourse and an artist’s recognition. This research provides a method for investigating Wikipedia profiles of painters using NLP attributes. Among unsupervised machine learning algorithms, the K-means is adopted to group the painters using the driven attributes from the content details of their profile pages. The clustering results are evaluated through a benchmark painter list and a qualitative review. The model findings reveal that the suggested approach effectively clusters the presented benchmark painter profiles, highlighting the potential of textual data analysis on painter profile similarities.
Download

Area 5 - Complexity in Risk and Predictive Modeling

Full Papers
Paper Nr: 27
Title:

Univariate GARCH Model for Futures Option Pricing: Application to Silver Mini Futures in Indian Commodity Market
Authors:

S Sapna and Biju R. Mohan
Abstract: This research investigates the pricing of options related to silver commodity futures within the Indian market, employing a standard univariate Generalized Autoregressive Conditional Heteroscedastic (GARCH) model with a symmetric normal distribution for return modelling. The study evaluates the performance of this option pricing model specifically for silver mini futures options traded on the Multi Commodity Exchange. Furthermore, it compares the option prices determined using the GARCH model parameters with those calculated using the Black-76 model. The findings demonstrate that the option prices derived from the GARCH model fall consistently within the bid-ask price range and significantly outperform the Black-76 model in terms of option pricing accuracy. This underscores the practical utility of GARCH models in the context of the Indian commodity market. To the best of our knowledge, this research marks the pioneering attempt to incorporate parameters generated by the GARCH model for futures option pricing within the Indian commodity market.
Download
Paper Nr: 29
Title:

Harnessing Convolution and PageRank in Graph-Based Models for Enhanced Stock Selection Amid Market Volatility
Authors:

Siu T. Wong, Pin Ni, Francesca Medda and Victor Chang
Abstract: In a financial landscape marked by the aftermath of the pandemic and the inherent volatility of today’s markets, investors and fund managers grapple with the challenge of selecting stocks that can withstand short-term fluctuations and yield long-term gains. Traditional strategies, deeply rooted in a company’s financial metrics, confront the capricious nature of contemporary markets that are heavily influenced by a myriad of fast-changing factors. Recognizing the need for tools that can navigate through the confluence of these complex conditions, this paper introduces a graph-based approach. Employing a single-layered convolution method applied to a curated list of stocks, the study focuses on crafting a Graph-and-PageRank-based ranking system that directs investors to the most promising stocks. The approach leverages volatility and relative market capitalization as pivotal factors to gauge the interconnected influence of stocks, embodying the multifaceted decision-making process akin to that of a seasoned investor. This methodology not only aids in mitigating risks in the short term but also sets the stage for potential long-term investment strategies, paving the way for robust financial modeling that is both responsive to market dynamics and intuitive to the investor’s stock-picking rationale.
Paper Nr: 30
Title:

Modeling Networks of Interdependent Infrastructure in Complex Urban Environments Using Open-Data
Authors:

Antonio Di Pietro, Francesco Cavedon, Vittorio Rosato and George Stergiopoulos
Abstract: Dependency effects between Critical Infrastructure (CI) elements represent key information needed to predict and analyze the impact of natural (or man-made) disturbances. The dependency links among CI elements and their associated weight are data whose availability is often complex to determine and are usually not available. Leveraging on several data supporting US and EU Directives for the Resilience and the Protection of CIs, the objective of the present work is to define a dependency network of elements of critical sectors extracted from available open-data. The resulting network is then studied in terms of its basic topological properties. The analysis of the network provides interesting clues about the properties and locations of critical points that can cause cascading failures. In addition, this information can form the basis for planning actions that mitigate the risk of cascading effects.
Download
Short Papers
Paper Nr: 11
Title:

Outage Risks: It is not the Malicious Attacks that Take Down Your Service
Authors:

Jan M. Evang and Alojz Gomola
Abstract: Network operators often prioritize risks to optimize resource allocation. This paper introduces an analysis model for prioritizing network outage risks, a practice common among large operators but underrepresented in research. We propose metrics such as Risk Value and clarify their definitions. The study reveals insights into the impact of incidents, classifying them based on customer support cases. Through the application of the presented method to a global network operator, we demonstrate the generation of unexpected insights and outcomes: Short outages are very frequent and regularly cause customer complaints. We examined both safety and security related incidents in relationship to customer report events, with surprising turn out for malicious attacks.
Download