Athanasios Tsakalidis
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Athanasios Tsakalidis has a total of 34 co-authors. Here are the top co-authors by number of shared publications:
- Spyros Sioutas
Ionian University
6 shared publications - Andreas Kanavos
Department of Informatics, Ionian University, 49100 Corfu, Greece
4 shared publications - Giannis Tzimas
Data and Media Laboratory, Department of Electrical and Computer Engineering, University of Peloponnese, 26334 Patras, Greece
4 shared publications - Evangelos Sakkopoulos
Department of Informatics, University of Piraeus, Piraeus, Greece
3 shared publications - Nikolaos Nodarakis
University of Patras
3 shared publications - Christos Makris
University of Patras
2 shared publications
Latest Publications
- Open Access
- 28 July 2020
- Published by MDPI AG in Algorithms
- Vol. 13 (8) , 182
Trajectory Clustering and k-NN for Robust Privacy Preserving k-NN Query Processing in GeoSpark
Privacy Preserving and Anonymity have gained significant concern from the big data perspective. We have the view that the forthcoming frameworks and theories will establish several solutions for privacy protection. The k-anonymity is considered a key solution that has been widely employed to prevent data re-identifcation and concerns us in the context of this work. Data modeling has also gained significant attention from the big data perspective. It is believed that the advancing distributed environments will provide users with several solutions for efficient spatio-temporal data management. GeoSpark will be utilized in the current work as it is a key solution that has been widely employed for spatial data. Specifically, it works on the top of Apache Spark, the main framework leveraged from the research community and organizations for big data transformation, processing and visualization. To this end, we focused on trajectory data representation so as to be applicable to the GeoSpark environment, and a GeoSpark-based approach is designed for the efficient management of real spatio-temporal data. Th next step is to gain deeper understanding of the data through the application of k nearest neighbor (k-NN) queries either using indexing methods or otherwise. The k-anonymity set computation, which is the main component for privacy preservation evaluation and the main issue of our previous works, is evaluated in the GeoSpark environment. More to the point, the focus here is on the time cost of k-anonymity set computation along with vulnerability measurement. The extracted results are presented into tables and figures for visual inspection.
- 28 July 2020
- Published by MDPI AG in Algorithms
- Vol. 13 (8) , 182
Open AccessTrajectory Clustering and k-NN for Robust Privacy Preserving k-NN Query Processing in GeoSpark
Privacy Preserving and Anonymity have gained significant concern from the big data perspective. We have the view that the forthcoming frameworks and theories will establish several solutions for privacy protection. The k-anonymity is considered a key solution that has been widely employed to prevent data re-identifcation and concerns us in the context of this work. Data modeling has also gained significant attention from the big data perspective. It is believed that the advancing distributed environments will provide users with several solutions for efficient spatio-temporal data management. GeoSpark will be utilized in the current work as it is a key solution that has been widely employed for spatial data. Specifically, it works on the top of Apache Spark, the main framework leveraged from the research community and organizations for big data transformation, processing and visualization. To this end, we focused on trajectory data representation so as to be applicable to the GeoSpark environment, and a GeoSpark-based approach is designed for the efficient management of real spatio-temporal data. Th next step is to gain deeper understanding of the data through the application of k nearest neighbor (k-NN) queries either using indexing methods or otherwise. The k-anonymity set computation, which is the main component for privacy preservation evaluation and the main issue of our previous works, is evaluated in the GeoSpark environment. More to the point, the focus here is on the time cost of k-anonymity set computation along with vulnerability measurement. The extracted results are presented into tables and figures for visual inspection.
- Open Access
- 11 December 2019
- Published by MDPI AG in Algorithms
- Vol. 12 (12) , 266
Storage Efficient Trajectory Clustering and k-NN for Robust Privacy Preserving Spatio-Temporal Databases
The need to store massive volumes of spatio-temporal data has become a difficult task as GPS capabilities and wireless communication technologies have become prevalent to modern mobile devices. As a result, massive trajectory data are produced, incurring expensive costs for storage, transmission, as well as query processing. A number of algorithms for compressing trajectory data have been proposed in order to overcome these difficulties. These algorithms try to reduce the size of trajectory data, while preserving the quality of the information. In the context of this research work, we focus on both the privacy preservation and storage problem of spatio-temporal databases. To alleviate this issue, we propose an efficient framework for trajectories representation, entitled DUST (DUal-based Spatio-temporal Trajectory), by which a raw trajectory is split into a number of linear sub-trajectories which are subjected to dual transformation that formulates the representatives of each linear component of initial trajectory; thus, the compressed trajectory achieves compression ratio equal to . To our knowledge, we are the first to study and address k-NN queries on nonlinear moving object trajectories that are represented in dual dimensional space. Additionally, the proposed approach is expected to reinforce the privacy protection of such data. Specifically, even in case that an intruder has access to the dual points of trajectory data and try to reproduce the native points that fit a specific component of the initial trajectory, the identity of the mobile object will remain secure with high probability. In this way, the privacy of the k-anonymity method is reinforced. Through experiments on real spatial datasets, we evaluate the robustness of the new approach and compare it with the one studied in our previous work.- 11 December 2019
- Published by MDPI AG in Algorithms
- Vol. 12 (12) , 266
Open AccessStorage Efficient Trajectory Clustering and k-NN for Robust Privacy Preserving Spatio-Temporal Databases
The need to store massive volumes of spatio-temporal data has become a difficult task as GPS capabilities and wireless communication technologies have become prevalent to modern mobile devices. As a result, massive trajectory data are produced, incurring expensive costs for storage, transmission, as well as query processing. A number of algorithms for compressing trajectory data have been proposed in order to overcome these difficulties. These algorithms try to reduce the size of trajectory data, while preserving the quality of the information. In the context of this research work, we focus on both the privacy preservation and storage problem of spatio-temporal databases. To alleviate this issue, we propose an efficient framework for trajectories representation, entitled DUST (DUal-based Spatio-temporal Trajectory), by which a raw trajectory is split into a number of linear sub-trajectories which are subjected to dual transformation that formulates the representatives of each linear component of initial trajectory; thus, the compressed trajectory achieves compression ratio equal to . To our knowledge, we are the first to study and address k-NN queries on nonlinear moving object trajectories that are represented in dual dimensional space. Additionally, the proposed approach is expected to reinforce the privacy protection of such data. Specifically, even in case that an intruder has access to the dual points of trajectory data and try to reproduce the native points that fit a specific component of the initial trajectory, the identity of the mobile object will remain secure with high probability. In this way, the privacy of the k-anonymity method is reinforced. Through experiments on real spatial datasets, we evaluate the robustness of the new approach and compare it with the one studied in our previous work. - Open Access
- 3 April 2017
- Published by MDPI AG in Computation
- Vol. 5 (2) , 20
Detecting Perturbed Subpathways towards Mouse Lung Regeneration Following H1N1 Influenza Infection
It has already been established by the systems-level approaches that the future of predictive disease biomarkers will not be sketched by plain lists of genes or proteins or other biological entities but rather integrated entities that consider all underlying component relationships. Towards this orientation, early pathway-based approaches coupled expression data with whole pathway interaction topologies but it was the recent approaches that zoomed into subpathways (local areas of the entire biological pathway) that provided more targeted and context-specific candidate disease biomarkers. Here, we explore the application potential of PerSubs, a graph-based algorithm which identifies differentially activated disease-specific subpathways. PerSubs is applicable both for microarray and RNA-Seq data and utilizes the Kyoto Encyclopedia of Genes and Genomes (KEGG) database as reference for biological pathways. PerSubs operates in two stages: first, identifies differentially expressed genes (or uses any list of disease-related genes) and in second stage, treating each gene of the list as start point, it scans the pathway topology around to build meaningful subpathway topologies. Here, we apply PerSubs to investigate which pathways are perturbed towards mouse lung regeneration following H1N1 influenza infection.
- 3 April 2017
- Published by MDPI AG in Computation
- Vol. 5 (2) , 20
Open AccessDetecting Perturbed Subpathways towards Mouse Lung Regeneration Following H1N1 Influenza Infection
It has already been established by the systems-level approaches that the future of predictive disease biomarkers will not be sketched by plain lists of genes or proteins or other biological entities but rather integrated entities that consider all underlying component relationships. Towards this orientation, early pathway-based approaches coupled expression data with whole pathway interaction topologies but it was the recent approaches that zoomed into subpathways (local areas of the entire biological pathway) that provided more targeted and context-specific candidate disease biomarkers. Here, we explore the application potential of PerSubs, a graph-based algorithm which identifies differentially activated disease-specific subpathways. PerSubs is applicable both for microarray and RNA-Seq data and utilizes the Kyoto Encyclopedia of Genes and Genomes (KEGG) database as reference for biological pathways. PerSubs operates in two stages: first, identifies differentially expressed genes (or uses any list of disease-related genes) and in second stage, treating each gene of the list as start point, it scans the pathway topology around to build meaningful subpathway topologies. Here, we apply PerSubs to investigate which pathways are perturbed towards mouse lung regeneration following H1N1 influenza infection.
- Open Access
- 4 March 2017
- Published by MDPI AG in Algorithms
- Vol. 10 (1) , 33
Large Scale Implementations for Twitter Sentiment Classification
Sentiment Analysis on Twitter Data is indeed a challenging problem due to the nature, diversity and volume of the data. People tend to express their feelings freely, which makes Twitter an ideal source for accumulating a vast amount of opinions towards a wide spectrum of topics. This amount of information offers huge potential and can be harnessed to receive the sentiment tendency towards these topics. However, since no one can invest an infinite amount of time to read through these tweets, an automated decision making approach is necessary. Nevertheless, most existing solutions are limited in centralized environments only. Thus, they can only process at most a few thousand tweets. Such a sample is not representative in order to define the sentiment polarity towards a topic due to the massive number of tweets published daily. In this work, we develop two systems: the first in the MapReduce and the second in the Apache Spark framework for programming with Big Data. The algorithm exploits all hashtags and emoticons inside a tweet, as sentiment labels, and proceeds to a classification method of diverse sentiment types in a parallel and distributed manner. Moreover, the sentiment analysis tool is based on Machine Learning methodologies alongside Natural Language Processing techniques and utilizes Apache Spark’s Machine learning library, MLlib. In order to address the nature of Big Data, we introduce some pre-processing steps for achieving better results in Sentiment Analysis as well as Bloom filters to compact the storage size of intermediate data and boost the performance of our algorithm. Finally, the proposed system was trained and validated with real data crawled by Twitter, and, through an extensive experimental evaluation, we prove that our solution is efficient, robust and scalable while confirming the quality of our sentiment identification.
- 4 March 2017
- Published by MDPI AG in Algorithms
- Vol. 10 (1) , 33
Open AccessLarge Scale Implementations for Twitter Sentiment Classification
Sentiment Analysis on Twitter Data is indeed a challenging problem due to the nature, diversity and volume of the data. People tend to express their feelings freely, which makes Twitter an ideal source for accumulating a vast amount of opinions towards a wide spectrum of topics. This amount of information offers huge potential and can be harnessed to receive the sentiment tendency towards these topics. However, since no one can invest an infinite amount of time to read through these tweets, an automated decision making approach is necessary. Nevertheless, most existing solutions are limited in centralized environments only. Thus, they can only process at most a few thousand tweets. Such a sample is not representative in order to define the sentiment polarity towards a topic due to the massive number of tweets published daily. In this work, we develop two systems: the first in the MapReduce and the second in the Apache Spark framework for programming with Big Data. The algorithm exploits all hashtags and emoticons inside a tweet, as sentiment labels, and proceeds to a classification method of diverse sentiment types in a parallel and distributed manner. Moreover, the sentiment analysis tool is based on Machine Learning methodologies alongside Natural Language Processing techniques and utilizes Apache Spark’s Machine learning library, MLlib. In order to address the nature of Big Data, we introduce some pre-processing steps for achieving better results in Sentiment Analysis as well as Bloom filters to compact the storage size of intermediate data and boost the performance of our algorithm. Finally, the proposed system was trained and validated with real data crawled by Twitter, and, through an extensive experimental evaluation, we prove that our solution is efficient, robust and scalable while confirming the quality of our sentiment identification.
- Open Access
Using Hadoop for Large Scale Analysis on Twitter: A Technical Report
Sentiment analysis (or opinion mining) on Twitter data has attracted much attention recently. One of the system's key features, is the immediacy in communication with other users in an easy, user-friendly and fast way. Consequently, people tend to express their feelings freely, which makes Twitter an ideal source for accumulating a vast amount of opinions towards a wide diversity of topics. This amount of information offers huge potential and can be harnessed to receive the sentiment tendency towards these topics. However, since none can invest an infinite amount of time to read through these tweets, an automated decision making approach is necessary. Nevertheless, most existing solutions are limited in centralized environments only. Thus, they can only process at most a few thousand tweets. Such a sample, is not representative to define the sentiment polarity towards a topic due to the massive number of tweets published daily. In this paper, we go one step further and develop a novel method for sentiment learning in the MapReduce framework. Our algorithm exploits the hashtags and emoticons inside a tweet, as sentiment labels, and proceeds to a classification procedure of diverse sentiment types in a parallel and distributed manner. Moreover, we utilize Bloom filters to compact the storage size of intermediate data and boost the performance of our algorithm. Through an extensive experimental evaluation, we prove that our solution is efficient, robust and scalable and confirm the quality of our sentiment identification.
Open AccessUsing Hadoop for Large Scale Analysis on Twitter: A Technical Report
Sentiment analysis (or opinion mining) on Twitter data has attracted much attention recently. One of the system's key features, is the immediacy in communication with other users in an easy, user-friendly and fast way. Consequently, people tend to express their feelings freely, which makes Twitter an ideal source for accumulating a vast amount of opinions towards a wide diversity of topics. This amount of information offers huge potential and can be harnessed to receive the sentiment tendency towards these topics. However, since none can invest an infinite amount of time to read through these tweets, an automated decision making approach is necessary. Nevertheless, most existing solutions are limited in centralized environments only. Thus, they can only process at most a few thousand tweets. Such a sample, is not representative to define the sentiment polarity towards a topic due to the massive number of tweets published daily. In this paper, we go one step further and develop a novel method for sentiment learning in the MapReduce framework. Our algorithm exploits the hashtags and emoticons inside a tweet, as sentiment labels, and proceeds to a classification procedure of diverse sentiment types in a parallel and distributed manner. Moreover, we utilize Bloom filters to compact the storage size of intermediate data and boost the performance of our algorithm. Through an extensive experimental evaluation, we prove that our solution is efficient, robust and scalable and confirm the quality of our sentiment identification.
- 29 September 2015
- Published by Springer Nature
- p. 154-165
Data Engineering for Materials Identification, Damage Assessment and Restoration of Cultural Objects
- 29 September 2015
- Published by Springer Nature
- p. 154-165
Data Engineering for Materials Identification, Damage Assessment and Restoration of Cultural Objects
- 3 November 2014
- Published by Springer Nature
- Vol. 820, 79-89
Exploiting Expert Systems in Cardiology: A Comparative Study
An improved Adaptive Neuro-Fuzzy Inference System (ANFIS) in the field of critical cardiovascular diseases is presented. The system stems from an earlier application based only on a Sugeno-type Fuzzy Expert System (FES) with the addition of an Artificial Neural Network (ANN) computational structure. Thus, inherent characteristics of ANNs, along with the human-like knowledge representation of fuzzy systems are integrated. The ANFIS has been utilized into building five different sub-systems, distinctly covering Coronary Disease, Hypertension, Atrial Fibrillation, Heart Failure, and Diabetes, hence aiding doctors of medicine (MDs), guide trainees, and encourage medical experts in their diagnoses centering a wide range of Cardiology. The Fuzzy Rules have been trimmed down and the ANNs have been optimized in order to focus into each particular disease and produce results ready-to-be applied to real-world patients.
- 3 November 2014
- Published by Springer Nature
- Vol. 820, 79-89
Exploiting Expert Systems in Cardiology: A Comparative Study
An improved Adaptive Neuro-Fuzzy Inference System (ANFIS) in the field of critical cardiovascular diseases is presented. The system stems from an earlier application based only on a Sugeno-type Fuzzy Expert System (FES) with the addition of an Artificial Neural Network (ANN) computational structure. Thus, inherent characteristics of ANNs, along with the human-like knowledge representation of fuzzy systems are integrated. The ANFIS has been utilized into building five different sub-systems, distinctly covering Coronary Disease, Hypertension, Atrial Fibrillation, Heart Failure, and Diabetes, hence aiding doctors of medicine (MDs), guide trainees, and encourage medical experts in their diagnoses centering a wide range of Cardiology. The Fuzzy Rules have been trimmed down and the ANNs have been optimized in order to focus into each particular disease and produce results ready-to-be applied to real-world patients.
- Open Access
- 2 October 2013
- Published by Springer Nature in BMC Systems Biology
- Vol. 7 (1) , 96
Reconstruction of the experimentally supported human protein interactome: what can we learn?
Background: Understanding the topology and dynamics of the human protein-protein interaction (PPI) network will significantly contribute to biomedical research, therefore its systematic reconstruction is required. Several meta-databases integrate source PPI datasets, but the protein node sets of their networks vary depending on the PPI data combined. Due to this inherent heterogeneity, the way in which the human PPI network expands via multiple dataset integration has not been comprehensively analyzed. We aim at assembling the human interactome in a global structured way and exploring it to gain insights of biological relevance. Results: First, we defined the UniProtKB manually reviewed human “complete” proteome as the reference protein-node set and then we mined five major source PPI datasets for direct PPIs exclusively between the reference proteins. We updated the protein and publication identifiers and normalized all PPIs to the UniProt identifier level. The reconstructed interactome covers approximately 60% of the human proteome and has a scale-free structure. No apparent differentiating gene functional classification characteristics were identified for the unrepresented proteins. The source dataset integration augments the network mainly in PPIs. Polyubiquitin emerged as the highest-degree node, but the inclusion of most of its identified PPIs may be reconsidered. The high number (>300) of connections of the subsequent fifteen proteins correlates well with their essential biological role. According to the power-law network structure, the unrepresented proteins should mainly have up to four connections with equally poorly-connected interactors. Conclusions: Reconstructing the human interactome based on the a priori definition of the protein nodes enabled us to identify the currently included part of the human “complete” proteome, and discuss the role of the proteins within the network topology with respect to their function. As the network expansion has to comply with the scale-free theory, we suggest that the core of the human interactome has essentially emerged. Thus, it could be employed in systems biology and biomedical research, despite the considerable number of currently unrepresented proteins. The latter are probably involved in specialized physiological conditions, justifying the scarcity of related PPI information, and their identification can assist in designing relevant functional experiments and targeted text mining algorithms.
- 2 October 2013
- Published by Springer Nature in BMC Systems Biology
- Vol. 7 (1) , 96
Open AccessReconstruction of the experimentally supported human protein interactome: what can we learn?
Background: Understanding the topology and dynamics of the human protein-protein interaction (PPI) network will significantly contribute to biomedical research, therefore its systematic reconstruction is required. Several meta-databases integrate source PPI datasets, but the protein node sets of their networks vary depending on the PPI data combined. Due to this inherent heterogeneity, the way in which the human PPI network expands via multiple dataset integration has not been comprehensively analyzed. We aim at assembling the human interactome in a global structured way and exploring it to gain insights of biological relevance. Results: First, we defined the UniProtKB manually reviewed human “complete” proteome as the reference protein-node set and then we mined five major source PPI datasets for direct PPIs exclusively between the reference proteins. We updated the protein and publication identifiers and normalized all PPIs to the UniProt identifier level. The reconstructed interactome covers approximately 60% of the human proteome and has a scale-free structure. No apparent differentiating gene functional classification characteristics were identified for the unrepresented proteins. The source dataset integration augments the network mainly in PPIs. Polyubiquitin emerged as the highest-degree node, but the inclusion of most of its identified PPIs may be reconsidered. The high number (>300) of connections of the subsequent fifteen proteins correlates well with their essential biological role. According to the power-law network structure, the unrepresented proteins should mainly have up to four connections with equally poorly-connected interactors. Conclusions: Reconstructing the human interactome based on the a priori definition of the protein nodes enabled us to identify the currently included part of the human “complete” proteome, and discuss the role of the proteins within the network topology with respect to their function. As the network expansion has to comply with the scale-free theory, we suggest that the core of the human interactome has essentially emerged. Thus, it could be employed in systems biology and biomedical research, despite the considerable number of currently unrepresented proteins. The latter are probably involved in specialized physiological conditions, justifying the scarcity of related PPI information, and their identification can assist in designing relevant functional experiments and targeted text mining algorithms.
- 12 May 2013
- Published by Springer Nature in Journal of Medical Systems
- Vol. 37 (3) , 1-12
easyHealthApps: e-Health Apps Dynamic Generation for Smartphones & Tablets
Mobile phones and especially smartphones have been embraced by a rapidly increasing number of people worldwide and this trend is expected to evolve even more in the years to come. There are numerous smartphone Apps that record critical medical data in an effort to solve a particular health issue each time. We studied such applications and not surprisingly, we have found that development and design effort is often repeated. Software patterns have been detected to exist, however re-usability has not been enforced. This leads to lost programming manpower and to increased probability of repeating bugs in Apps. Moreover, at the moment smartphone e-Health Apps demand time, effort and costs for development. Unfortunately even simple data recording Apps are practically impossible to be produced by multiple health domain users who are not developers. In this work, we propose, design and implement a simple and integrated solution which gives healthcare professionals and researchers the ability to create their own data intensive smartphone applications, independent of the desired healthcare domain. The proposed approach applies efficient software techniques that hide development from the users and enable App creation through a simple Web User Interface. The Apps produced are in native format and it is possible to dynamically receive m-Health business logic and the chosen UI. Evaluation of the proposed solution has shown that the generated Apps are functionally and UI equivalent to human-coded Apps according to a number of comparison parameters. Furthermore, e-Health professionals show particular interest in developing Apps on their own for a particular domain they focus on.
- 12 May 2013
- Published by Springer Nature in Journal of Medical Systems
- Vol. 37 (3) , 1-12
easyHealthApps: e-Health Apps Dynamic Generation for Smartphones & Tablets
Mobile phones and especially smartphones have been embraced by a rapidly increasing number of people worldwide and this trend is expected to evolve even more in the years to come. There are numerous smartphone Apps that record critical medical data in an effort to solve a particular health issue each time. We studied such applications and not surprisingly, we have found that development and design effort is often repeated. Software patterns have been detected to exist, however re-usability has not been enforced. This leads to lost programming manpower and to increased probability of repeating bugs in Apps. Moreover, at the moment smartphone e-Health Apps demand time, effort and costs for development. Unfortunately even simple data recording Apps are practically impossible to be produced by multiple health domain users who are not developers. In this work, we propose, design and implement a simple and integrated solution which gives healthcare professionals and researchers the ability to create their own data intensive smartphone applications, independent of the desired healthcare domain. The proposed approach applies efficient software techniques that hide development from the users and enable App creation through a simple Web User Interface. The Apps produced are in native format and it is possible to dynamically receive m-Health business logic and the chosen UI. Evaluation of the proposed solution has shown that the generated Apps are functionally and UI equivalent to human-coded Apps according to a number of comparison parameters. Furthermore, e-Health professionals show particular interest in developing Apps on their own for a particular domain they focus on.