COMPLEXIS 2022 Abstracts


Area 1 - Complexity in Biology and Biomedical Engineering

Short Papers
Paper Nr: 15
Title:

Machine Learning Techniques for Breast Cancer Detection

Authors:

Karl Hall, Victor Chang and Paul Mitchell

Abstract: Breast cancer is the second most prevalent type of cancer overall and the most frequently occurring cancer in women. The most effective way to improve breast cancer survival rates still lies in the early detection of the disease. An increasingly popular and effective way of doing this is by using machine learning to classify and analyze patient data to help identify signs of cancer. This paper explores a variety of machine learning techniques and compares their prediction accuracy and other metrics when using the Breast Cancer Wisconsin (Original) data set using 10-fold cross-validation methods. Of the algorithms tested in this paper, a support vector machine model using the radial basis function kernel outperformed all other models we tested and those previously developed by others, achieving an accuracy of 99%.
Download

Area 2 - Complexity in Social Sciences

Full Papers
Paper Nr: 6
Title:

Automated Narratives: On the Influence of Bots in Narratives during the 2020 Vienna Terror Attack

Authors:

Lisa Grobelscheg, Ema Kušen and Mark Strembeck

Abstract: A narrative is a set of topic-wise interconnected messages that have been sent/posted via a social media platform. In recent years, social media play an important role in human information seeking behavior during and shortly after crisis events. Moreover, automated accounts (so called social bots) have been identified to play an instrumental role in manipulating the public discourse on social media. In this paper, we investigate the impact of bot accounts on the Twitter discourse surrounding the terror attack that took place in Vienna, Austria, on November 2nd 2020. The corresponding data-set consists of 399,247 tweets. In our analysis, we derive a structural topic model and map it to the five “narratives of crisis” as proposed by Seeger and Sellnow. Among other things, we were able to identify bot activity in neutral as well as in negative narratives, including breaking news updates, finger pointing, and expressions of shock and grief. Positive narratives, such as stories of heroes, were predominantly driven by human users. In addition, we found that the bots contributing to narratives surrounding the Vienna terror attack did not have the ability of picking up local story lines and contributed to more global narratives instead. Moreover, we identified similar temporal patterns in narratives with high bot involvement.
Download

Paper Nr: 10
Title:

Dynamics of Personal Responses to Terror Attacks: A Temporal Network Analysis Perspective

Authors:

Ema Kušen and Mark Strembeck

Abstract: In this paper, we analyze responses to terror attacks through the lens of the Terror Management Theory. We focus on the temporal evolution of Twitter messages that convey death anxiety, emotional pain, as well as positivity. We model the responses to terror attacks as personal reactions that include the use of a first person singular pronoun along with cues of affect and personal concerns. In order to detect these textual features, we used the Linguistic Inquiry and Word Count (LIWC) tool. Our data-set includes tweets related to three terror attacks: the 2017 Manchester terror attack, the 2019 Christchurch terror attack, and the 2020 Vienna terror attack. Our analysis is based on 3.8 million tweets that have been sent by 1.6 million users. The results indicate that positive messages associated with the use of religious words (e.g., messages of prayers and hope) dominate over those that convey emotional pain and fear of death. This points to a tendency to spread hope and empathy in the aftermath of a terror attack. We found that the acute phase of a terror attack exhibits a high volume of messages that sharply decline in the immediate aftermath. In contrast, positive messages exhibit smaller peaks even one week after a terror attack happened.
Download

Short Papers
Paper Nr: 1
Title:

Systematic Review on Cybersecurity Risks and Behaviours: Methodological Approaches

Authors:

Eliza Oliveira and Vania Baldi

Abstract: This paper presents a systematic literature review regarding risk perception and precautionary behaviour related to cybersecurity. Objectives encompassed identifying issues related to methodological approaches, the studies’ operationalisation and other essential topics, highlighting significant gaps to be fulfilled in future investigations. The study included a search in the multidisciplinary databases of Science Direct, Web of Science, and the Scientific Repositories of Open Access in Portugal, focusing on publications after 2016. A total of nine articles were analysed. The review was developed using the Preferred Reporting Items for Systematic Review and Meta-analysis Protocols’ research method. Also, publications were coded using the NVivo 12 software for synthesising the results. The small number of studies considered for analysis revealed that risk perception and precautionary behaviour concerning cybersecurity is still an under-explored study area. Furthermore, methodological gaps are highlighted for future works. Studies in the cybersecurity field provide a dataset for policymakers, directing efforts and predicting responses to digital technologies, making this subject matter highly substantial.
Download

Paper Nr: 14
Title:

Sustainability and Goal Fitness Index for the Analysis of Sustainable Development Goals: A Methodological Proposal

Authors:

Sanny González, Gabriel Pereira and Arturo González

Abstract: The Sustainable Development Goals (SDGs) were adopted in September 2015 by the 193 member states of the United Nations (UN), which include 17 goals, 169 targets and 244 indicators, as an attempt to radically change the approach of the Sustainable Development Goals. Millennium Development (MDG). Since the adoption of the 2030 Agenda, the scientific community has increased its interest in the evaluation, analysis, and evaluation of the interrelationships between the SDGs, proposing different approaches and using a diversity of methodological tools for the interactions of the SDGs. This research proposes a methodology that takes advantage of the concepts of Economic Fitness for the creation of a Sustainability Fitness Index (SFI) for the countries and a Goal Fitness Index (GFI) for each SDG. These indices are intended to provide a tool to analyze the interrelationships between the Sustainable Development Goals in such a way that they offer a new approach to address the capacities of the countries and the fulfillment of the SDGs. The results of the SFI are a first attempt to identify development priorities aligned with the SDGs in each country, based on their available productive capacities, which could help make more efficient use of their limited resources and increase the achievement of the SDGs.
Download

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

Short Papers
Paper Nr: 7
Title:

CLIP Augmentation for Image Search

Authors:

Ingus J. Pretkalnins, Arturs Sprogis and Guntis Barzdins

Abstract: We devised a probabilistic method for adding face recognition to the neural network model CLIP. The method was tested by creating a prototype and matching 1000 images to their descriptions. The method improved the text to image Recall @ 1 metric from 14.0% matches for CLIP alone to 21.8% for CLIP + method, for a sample size of 1000 images and descriptions.
Download

Paper Nr: 12
Title:

Moving Other Way: Exploring Word Mover Distance Extensions

Authors:

Ilya S. Smirnov and Ivan P. Yamshchikov

Abstract: The word mover’s distance (WMD) is a popular semantic similarity metric for two documents. This metric is quite interpretable and reflects the similarity well, but some aspects can be improved. This position paper studies several possible extensions of WMD. We introduce some regularizations of WMD based on a word match and the frequency of words in the corpus as a weighting factor. Besides, we calculate WMD in word vector spaces with non-Euclidean geometry and compare it with the metric in Euclidean space. We validate possible extensions of WMD on six document classification datasets. Some proposed extensions show better results in terms of the k-nearest neighbor classification error than WMD.
Download

Area 4 - Complexity in Informatics and Networking

Short Papers
Paper Nr: 3
Title:

Consultation to Effectiveness and Logical Meaning

Authors:

Susumu Yamasaki and Mariko Sasakura

Abstract: This paper deals with consultation as a role of consultant like a teacher in a virtual reality class for lessons. The consultation subject is taken in causal analysis and database design to conceive causal relation. The causal analysis is abstracted into effectiveness which is caused by requisites and prohibitions for effects. The database design is represented as recursive Backus-Naur Form for structural expressions. Then abstract consultation can be regarded as a function to transform effectiveness into recursive representation. The sense or meaning of such abstract consultation may be a semantics of conditioned formulas in modal logic. The modal logic is formulated in this paper, by extending postfix modality for name of database design as well as greatest fixed point operator for denoting an equilibrium state set where the transformation of effectiveness with prefix modality to design with postfix modality may be available. The conditioning of formulas is given by a state set so that the formula may be considered in terms of propositional base. As regards recursive representations in Backus-Naur Form for database design, model theory is established in 3-valued domain, for the sense of strong negation in accordance to prohibitions influencing effectiveness. Some fixed point is taken for the mapping associated with recursive representations, where the mapping is related to formations of models over 3-valued domain. The fixed point model can be seen as means of retrieval. However, because the mapping is generally a nonmonotonic function whose fixed point semantics is not always available, we have another method to adopt negation as failure for strong negation such that the model of a given recursive representation as database is always guaranteed if the representation takes a restricted form. Abstract consultation to effectiveness for database design is thus formulated, with model theory in database representations.
Download

Paper Nr: 11
Title:

Extracting Mass Transportation Networks from General Transit Feed Specification Datasets

Authors:

Gergely Kocsis and Imre Varga

Abstract: In several smart-city applications the networks of the mass-transportation systems can be bases of investigations. In this paper we show how one can extract a network of connected stops from the General Transit Feed Specification (GTFS) feed of a given service provider. We have also implemented this process as a tool (gtfs2net) that is available for use at the GitHub page of the project. On of our most important finding is that since providers do not follow the specification in a coherent way regarding the use of parent stations the problem of close stops has to be manually handled. In order to show how our tool works in practice we have provided some extracted networks with their properties.
Download

Paper Nr: 4
Title:

The Hybrid Model of Broken Agile Transformation in Big Telco Corporations

Authors:

Dragan Stankovski, Tsvyatko D. Bikov and Dimitar I. Radev

Abstract: The “Technical depth” it's one of the major challenges leading most of the project to delays or failures. Implementation of the Agile approach in its pure form does not fit do needs of the big corporations providing services in the telecommunications branch. This paper aims to present a hybrid model of “Broken Agile” that will accommodate and increase with a significant level the software delivery and development. The approach is resolving and providing a formula for the improvement of already working solutions in Agile projects.
Download

Area 5 - Complexity in Risk and Predictive Modeling

Full Papers
Paper Nr: 8
Title:

Non-linear Black-Scholes Option Pricing Model based on Quantum Dynamics

Authors:

Marcin Wróblewski and Andrzej Myślinski

Abstract: This paper is concerned with the option pricing based on an extended non-linear Black-Scholes model. In literature, non-linear Black-Scholes models are formulated assuming the stochastic asset price volatility, volatile risk-free interest rate or the occurrence of the transaction costs. Since these assumptions are matching better the real market conditions, these models are regarded to be more accurate in option pricing than linear Black- Scholes model. In this paper the option pricing model is derived from non-linear Schrödinger equation. This equation governs the movement of quantum particles which is similar to the volatility of stock prices. The nonlinear Schrödinger equation with external potential terms is formulated. The non-linear Black-Scholes option pricing model is formulated using the transformation of the non-linear Schrödinger equation from complex Hilbert to real Euclidean space. The developed model has been used to predict European call options price based on WIG20 stock prices. The model parameters have been estimated based on real market data. The method of lines has been used to solve numerically the non-linear option pricing model. The model parameters have been estimated based on real market data. Numerical results are provided and discussed. The obtained results confirm high accuracy of the proposed non-linear model.
Download

Short Papers
Paper Nr: 13
Title:

Attention! Transformer with Sentiment on Cryptocurrencies Price Prediction

Authors:

Huali Zhao, Martin Crane and Marija Bezbradica

Abstract: Cryptocurrencies have won a lot of attention as an investment tool in recent years. Specific research has been done on cryptocurrencies’ price prediction while the prices surge up. Classic models and recurrent neural networks are applied for the time series forecast. However, there remains limited research on how the Transformer works on forecasting cryptocurrencies price data. This paper investigated the forecasting capability of the Transformer model on Bitcoin (BTC) price data and Ethereum (ETH) price data which are time series with high fluctuation. Long short term memory model (LSTM) is employed for performance comparison. The result shows that LSTM performs better than Transformer both on BTC and ETH price prediction. Furthermore, in this paper, we also investigated if sentiment analysis can help improve the model’s performance in forecasting future prices. Twitter data and Valence Aware Dictionary and sEntiment Reasoner (VADER) is used for getting sentiment scores. The result shows that the sentiment analysis improves the Transformer model’s performance on BTC price but not ETH price. For the LSTM model, the sentiment analysis does not help with prediction results. Finally, this paper also shows that transfer learning can help on improving the Transformer’s prediction ability on ETH price data.
Download

Paper Nr: 9
Title:

A Computational Pipeline for Modeling and Predicting Wildfire Behavior

Authors:

Nuno Fachada

Abstract: Wildfires constitute a major socioeconomic burden. While a number of scientific and technological methods have been used for predicting and mitigating wildfires, this is still an open problem. In turn, agent-based modeling is a modeling approach where each entity of the system being modeled is represented as an independent decision-making agent. It is a useful technique for studying systems that can be modeled in terms of interactions between individual components. Consequently, it is an interesting methodology for modeling wildfire behavior. In this position paper, we propose a complete computational pipeline for modeling and predicting wildfire behavior by leveraging agent-based modeling, among other techniques. This project is to be developed in collaboration with scientific and civil stakeholders, and should produce an open decision support system easily extendable by stakeholders and other interested parties.
Download