Panagiotis Stamatopoulos
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Panagiotis Stamatopoulos has a total of 31 co-authors. Here are the top co-authors by number of shared publications:
Nikolaos Pothitos
Department of Informatics and Telecommunications, National and Kapodistrian University of Athens, Panepistimiopolis, 157 84 Athens, Greece
5 shared publicationsVangelis Karkaletsis
National Centre for Scientific Research Demokritos, Demokritos, Greece
4 shared publicationsGeorge Boukeas
University of Athens
3 shared publicationsKyriakos Zervoudakis
3 shared publicationsConstantin Halatsis
University of Athens
2 shared publicationsConstantinos Halatsis
University of Athens
2 shared publications
Latest Publications
- Open Access
- 17 July 2023
- Published by Association for Computing Machinery (ACM) in ACM Computing Surveys
- Vol. 55 (14s) , 1-40
Content-based and Knowledge-enriched Representations for Classification Across Modalities: A Survey
This survey documents representation approaches for classification across different modalities, from purely content-based methods to techniques utilizing external sources of structured knowledge. We present studies related to three paradigms used for representation, namely (a) low-level template-matching methods, (b) aggregation-based approaches, and (c) deep representation learning systems. We then describe existing resources of structure knowledge and elaborate on the need for enriching representations with such information. Approaches that utilize knowledge resources are presented next, organized with respect to how external information is exploited, i.e., (a) input enrichment and modification, (b) knowledge-based refinement and (c) end-to-end knowledge-aware systems. We subsequently provide a high-level discussion to summarize and compare strengths/weaknesses of the representation/enrichment paradigms proposed, and conclude the survey with an overview of relevant research findings and possible directions for future work.
- 17 July 2023
- Published by Association for Computing Machinery (ACM) in ACM Computing Surveys
- Vol. 55 (14s) , 1-40
Open AccessContent-based and Knowledge-enriched Representations for Classification Across Modalities: A Survey
This survey documents representation approaches for classification across different modalities, from purely content-based methods to techniques utilizing external sources of structured knowledge. We present studies related to three paradigms used for representation, namely (a) low-level template-matching methods, (b) aggregation-based approaches, and (c) deep representation learning systems. We then describe existing resources of structure knowledge and elaborate on the need for enriching representations with such information. Approaches that utilize knowledge resources are presented next, organized with respect to how external information is exploited, i.e., (a) input enrichment and modification, (b) knowledge-based refinement and (c) end-to-end knowledge-aware systems. We subsequently provide a high-level discussion to summarize and compare strengths/weaknesses of the representation/enrichment paradigms proposed, and conclude the survey with an overview of relevant research findings and possible directions for future work.
- Open Access
- 18 May 2016
- Published by Association for Computing Machinery (ACM)
- p. 1-8
Piece of Pie Search
Search is not a direct path to a solution. While searching for a solution to a problem, heuristics consult us to avoid paths with dead ends, but they are not infallible. Many popular search methodologies \"disobey\" them during critical points of the search. In this work, we found an efficient stochastic methods framework that smoothly combines randomness with normal heuristics. We consider a factor of disobedience to the heuristics and we fine-tune it each time, according to our estimation of heuristic-reliability. We prove mathematically that while the disobedience factor falls, the stochastic methods approximate deterministic methods. Our algebraic evidence is supported by empirical evaluations on real life problems, such as course scheduling and frequency assignment. In this context, we exploit our proposed heuristic-reliability semantics in order to produce a piece of pie search (PoPS) method that can outperform other known constructive search processes in hard optimization problems.
- 18 May 2016
- Published by Association for Computing Machinery (ACM)
- p. 1-8
Open AccessPiece of Pie Search
Search is not a direct path to a solution. While searching for a solution to a problem, heuristics consult us to avoid paths with dead ends, but they are not infallible. Many popular search methodologies \"disobey\" them during critical points of the search. In this work, we found an efficient stochastic methods framework that smoothly combines randomness with normal heuristics. We consider a factor of disobedience to the heuristics and we fine-tune it each time, according to our estimation of heuristic-reliability. We prove mathematically that while the disobedience factor falls, the stochastic methods approximate deterministic methods. Our algebraic evidence is supported by empirical evaluations on real life problems, such as course scheduling and frequency assignment. In this context, we exploit our proposed heuristic-reliability semantics in order to produce a piece of pie search (PoPS) method that can outperform other known constructive search processes in hard optimization problems.
- Open Access
- 18 May 2016
- Published by Association for Computing Machinery (ACM)
- p. 1-4
Constraint Programming MapReduce'd
While Constraint Programming (CP) aims to explore efficiently large search trees, MapReduce (MR) is a framework that focuses on huge databases and text files. In this work, we try to bridge these two cutting-edge paradigms in order to solve Constraint Satisfaction Problems (CSPs) in distributed and/or parallel environments. We implement a CP system that is responsible to distribute the parts of the serial search tree to the workers in order to explore it. MR is responsible to allocate the physical resources of a cluster and assigns them to the workers. The CP and MR phases interchange and this novel coordination results into significant speedups while solving CSPs. Empirically, our approach is applied on common CSPs, such as N Queens and Number Partitioning.
- 18 May 2016
- Published by Association for Computing Machinery (ACM)
- p. 1-4
Open AccessConstraint Programming MapReduce'd
While Constraint Programming (CP) aims to explore efficiently large search trees, MapReduce (MR) is a framework that focuses on huge databases and text files. In this work, we try to bridge these two cutting-edge paradigms in order to solve Constraint Satisfaction Problems (CSPs) in distributed and/or parallel environments. We implement a CP system that is responsible to distribute the parts of the serial search tree to the workers in order to explore it. MR is responsible to allocate the physical resources of a cluster and assigns them to the workers. The CP and MR phases interchange and this novel coordination results into significant speedups while solving CSPs. Empirically, our approach is applied on common CSPs, such as N Queens and Number Partitioning.
- 1 November 2012
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2079 (10823409) , 335-343
Course Scheduling in an Adjustable Constraint Propagation Schema
Constraint Programming constitutes a prominent paradigm for solving time-consuming Constraint Satisfaction Problems (CSPs). In this work, at first we model a generic course scheduling problem as a CSP, that complies with the International Timetabling Competition (ITC) standards. Constraint Programming allowed us to search for a solution via several state-of-the-art methodologies and compare them. For the stochastic search methods, we propose new hybrid semi-random heuristics. Second, we chose to maintain bounds consistency during search to prune 'no-good' branches of the search tree. We theoretically define new lightweight consistency types, namely k-bounds-consistency, in order to speed up the overall search procedure. Eventually, we process real world data and show the efficiency of our proposal: While plain backtracking produces poor results, constraint propagation dramatically boosts the solutions quality, and can be 'fine-tuned' in our adjustable schema to make it even faster.
- 1 November 2012
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2079 (10823409) , 335-343
Course Scheduling in an Adjustable Constraint Propagation Schema
Constraint Programming constitutes a prominent paradigm for solving time-consuming Constraint Satisfaction Problems (CSPs). In this work, at first we model a generic course scheduling problem as a CSP, that complies with the International Timetabling Competition (ITC) standards. Constraint Programming allowed us to search for a solution via several state-of-the-art methodologies and compare them. For the stochastic search methods, we propose new hybrid semi-random heuristics. Second, we chose to maintain bounds consistency during search to prune 'no-good' branches of the search tree. We theoretically define new lightweight consistency types, namely k-bounds-consistency, in order to speed up the overall search procedure. Eventually, we process real world data and show the efficiency of our proposal: While plain backtracking produces poor results, constraint propagation dramatically boosts the solutions quality, and can be 'fine-tuned' in our adjustable schema to make it even faster.
- 1 January 2012
- Published by Springer Nature in Lecture Notes in Computer Science
- p. 9-16
Constraint Propagation as the Core of Local Search
- 1 January 2012
- Published by Springer Nature in Lecture Notes in Computer Science
- p. 9-16
Constraint Propagation as the Core of Local Search
- 1 December 2010
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 112-117
Expert based prediction of user preferences
Two influential strands in Recommender systems (RS) are the collaborative filtering and content based filtering that by taking into account user communities or interaction history suggest to the active user interesting items. However, the aforementioned approaches do not work well when confronted with new users with few interactions; or with the addition of new items. In such cases, the guidance of an expert could help the active user. In this paper we provide a definition of expert users that can be reduced into two components the expertise and the contribution. The former is related to the content of items evaluated by an expert and the latter refers to the influence of the expert to the users of a RS. In particular, contribution is learnt with the aid of a perceptron. Experts users are defined for values of the features of the items. Furthermore, we have studied the temporal evolution of the experts, as new users, new items, or new item evaluations are added into the system. Moreover, we have compared the proposed expert based method with a stereotype based method, since for both methods a minimal interaction of the active user with the RS suffices. The data originated from the MovieLens set with enhancements from the IMDB.
- 1 December 2010
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 112-117
Expert based prediction of user preferences
Two influential strands in Recommender systems (RS) are the collaborative filtering and content based filtering that by taking into account user communities or interaction history suggest to the active user interesting items. However, the aforementioned approaches do not work well when confronted with new users with few interactions; or with the addition of new items. In such cases, the guidance of an expert could help the active user. In this paper we provide a definition of expert users that can be reduced into two components the expertise and the contribution. The former is related to the content of items evaluated by an expert and the latter refers to the influence of the expert to the users of a RS. In particular, contribution is learnt with the aid of a perceptron. Experts users are defined for values of the features of the items. Furthermore, we have studied the temporal evolution of the experts, as new users, new items, or new item evaluations are added into the system. Moreover, we have compared the proposed expert based method with a stereotype based method, since for both methods a minimal interaction of the active user with the RS suffices. The data originated from the MovieLens set with enhancements from the IMDB.
- 1 January 2010
- Published by Springer Nature in Lecture Notes in Computer Science
- p. 405-410
Flexible Management of Large-Scale Integer Domains in CSPs
- 1 January 2010
- Published by Springer Nature in Lecture Notes in Computer Science
- p. 405-410
Flexible Management of Large-Scale Integer Domains in CSPs
- Open Access
- 13 October 2008
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Speech and Language Processing
- Vol. 5 (3) , 1-39
Summarization system evaluation revisited
This article presents a novel automatic method (AutoSummENG) for the evaluation of summarization systems, based on comparing the character n-gram graphs representation of the extracted summaries and a number of model summaries. The presented approach is language neutral, due to its statistical nature, and appears to hold a level of evaluation performance that matches and even exceeds other contemporary evaluation methods. Within this study, we measure the effectiveness of different representation methods, namely, word and character n-gram graph and histogram, different n-gram neighborhood indication methods as well as different comparison methods between the supplied representations. A theory for the a priori determination of the methods' parameters along with supporting experiments concludes the study to provide a complete alternative to existing methods concerning the automatic summary system evaluation process.
- 13 October 2008
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Speech and Language Processing
- Vol. 5 (3) , 1-39
Open AccessSummarization system evaluation revisited
This article presents a novel automatic method (AutoSummENG) for the evaluation of summarization systems, based on comparing the character n-gram graphs representation of the extracted summaries and a number of model summaries. The presented approach is language neutral, due to its statistical nature, and appears to hold a level of evaluation performance that matches and even exceeds other contemporary evaluation methods. Within this study, we measure the effectiveness of different representation methods, namely, word and character n-gram graph and histogram, different n-gram neighborhood indication methods as well as different comparison methods between the supplied representations. A theory for the a priori determination of the methods' parameters along with supporting experiments concludes the study to provide a complete alternative to existing methods concerning the automatic summary system evaluation process.
- 14 March 2007
- Published by Springer Nature in Journal of Intelligent Information Systems
- Vol. 30 (3) , 183-226
Using synchronic and diachronic relations for summarizing multiple documents describing evolving events
- 14 March 2007
- Published by Springer Nature in Journal of Intelligent Information Systems
- Vol. 30 (3) , 183-226
Using synchronic and diachronic relations for summarizing multiple documents describing evolving events
- 1 February 2005
- Published by Elsevier in Artificial Intelligence in Medicine
- Vol. 33 (2) , 157-177
Summarization from medical documents: a survey
- 1 February 2005
- Published by Elsevier in Artificial Intelligence in Medicine
- Vol. 33 (2) , 157-177
Summarization from medical documents: a survey