Francesca Da Ros
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Francesca Da Ros has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
- Luca Di Gaspero
University of Udine Italy
17 shared publications - Marie-Louise Lackner
Christian Doppler Laboratory for Artificial Intelligence and Optimization for Planning and Scheduling, DBAI, TU Wien, Vienna, Austria
7 shared publications - Nysret Musliu
Christian Doppler Laboratory for Artificial Intelligence and Optimization for Planning and Scheduling, DBAI, TU Wien, Vienna, Austria
7 shared publications - Kevin Roitero
University of Udine, udine, Italy
6 shared publications - Sara Ceschia
Polytechnic Department of Engineering and Architecture, University of Udine, Udine, Italy
4 shared publications - Andrea Schaerf
University of Udine, udine, Italy
3 shared publications
Latest Publications
- Open Access
- 31 October 2026
- Published by Springer Nature in Journal of Scheduling
- p. 1-20
Theoretical lower bounds for a parallel batch scheduling problem
The Oven Scheduling Problem (OSP) is an NP-hard real-world parallel batch scheduling problem arising in the semiconductor industry. Its objective is to schedule a set of jobs on ovens while minimizing total oven runtime, job tardiness, and setup costs. At the same time, it must adhere to various constraints such as oven eligibility and availability, job release dates, setup times between batches, and oven capacity limitations. In this paper, we develop theoretical, problem-specific lower bounds for the OSP that can be computed very quickly. We thoroughly examine these lower bounds, evaluating their quality and capability to improve existing solution methods. Moreover, these problem-specific lower bounds enable us to assess the solution quality for large instances for which exact methods often fail to provide tight lower bounds.
- 31 October 2026
- Published by Springer Nature in Journal of Scheduling
- p. 1-20
Open AccessTheoretical lower bounds for a parallel batch scheduling problem
The Oven Scheduling Problem (OSP) is an NP-hard real-world parallel batch scheduling problem arising in the semiconductor industry. Its objective is to schedule a set of jobs on ovens while minimizing total oven runtime, job tardiness, and setup costs. At the same time, it must adhere to various constraints such as oven eligibility and availability, job release dates, setup times between batches, and oven capacity limitations. In this paper, we develop theoretical, problem-specific lower bounds for the OSP that can be computed very quickly. We thoroughly examine these lower bounds, evaluating their quality and capability to improve existing solution methods. Moreover, these problem-specific lower bounds enable us to assess the solution quality for large instances for which exact methods often fail to provide tight lower bounds.
- Open Access
- 19 August 2026
- Published by Elsevier in Computers & Operations Research
- Vol. 196
Behavior and representation in open-weight Large Language Models for combinatorial optimization: From feature extraction to algorithm selection
- 19 August 2026
- Published by Elsevier in Computers & Operations Research
- Vol. 196
Open AccessBehavior and representation in open-weight Large Language Models for combinatorial optimization: From feature extraction to algorithm selection
- Open Access
- 7 August 2026
- Published by Elsevier in Computers & Operations Research
- Vol. 195
Revisiting local search for Curriculum-Based Course Timetabling
- 7 August 2026
- Published by Elsevier in Computers & Operations Research
- Vol. 195
Open AccessRevisiting local search for Curriculum-Based Course Timetabling
- Open Access
- 10 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 13-22
Assessing Reproducibility in Evolutionary Computation
Reproducibility is an important requirement in evolutionary computation, where results largely depend on computational experiments. In practice, reproducibility relies on how algorithms, experimental protocols, and artifacts are documented and shared. Despite growing awareness, there is still limited empirical evidence on the actual reproducibility levels of published work in the field. In this paper, we study the reproducibility practices in papers published in the Evolutionary Combinatorial Optimization and Metaheuristics track of the Genetic and Evolutionary Computation Conference over a ten-year period. We introduce a structured reproducibility checklist and apply it through a systematic manual assessment of the selected corpus. In addition, we propose RECAP (REproducibility Checklist Automation Pipeline), an LLM-based system that automatically evaluates reproducibility signals from paper text and associated code repositories. Our analysis shows that papers achieve an average completeness score of 0.62, and that 36.90% of them provide additional material beyond the manuscript itself. We demonstrate that automated assessment is feasible: RECAP achieves substantial agreement with human evaluators (Cohen's κ of 0.67). Together, these results highlight persistent gaps in reproducibility reporting and suggest that automated tools can effectively support large-scale, systematic monitoring of reproducibility practices.
- 10 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 13-22
Open AccessAssessing Reproducibility in Evolutionary Computation
Reproducibility is an important requirement in evolutionary computation, where results largely depend on computational experiments. In practice, reproducibility relies on how algorithms, experimental protocols, and artifacts are documented and shared. Despite growing awareness, there is still limited empirical evidence on the actual reproducibility levels of published work in the field. In this paper, we study the reproducibility practices in papers published in the Evolutionary Combinatorial Optimization and Metaheuristics track of the Genetic and Evolutionary Computation Conference over a ten-year period. We introduce a structured reproducibility checklist and apply it through a systematic manual assessment of the selected corpus. In addition, we propose RECAP (REproducibility Checklist Automation Pipeline), an LLM-based system that automatically evaluates reproducibility signals from paper text and associated code repositories. Our analysis shows that papers achieve an average completeness score of 0.62, and that 36.90% of them provide additional material beyond the manuscript itself. We demonstrate that automated assessment is feasible: RECAP achieves substantial agreement with human evaluators (Cohen's κ of 0.67). Together, these results highlight persistent gaps in reproducibility reporting and suggest that automated tools can effectively support large-scale, systematic monitoring of reproducibility practices.
- Top Cited PapersOpen Access
- 21 April 2026
- Published by Association for Computing Machinery (ACM) in ACM Computing Surveys
- Vol. 58 (11) , 1-53
Large Language Models for Combinatorial Optimization: A Systematic Review
This systematic review explores the application of Large Language Models (LLMs) in Combinatorial Optimization (CO). We report our findings using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. We conduct a literature search via Scopus and Google Scholar, examining over 2,000 publications. We assess publications against four inclusion and four exclusion criteria related to their language, research focus, publication year, and type. Eventually, we select 103 studies. We classify these studies into semantic categories and topics to provide a comprehensive overview of the field, including the tasks performed by LLMs, the architectures of LLMs, the existing datasets specifically designed for evaluating LLMs in CO, and the field of application. Finally, we identify future directions for leveraging LLMs in this field.
- 21 April 2026
- Published by Association for Computing Machinery (ACM) in ACM Computing Surveys
- Vol. 58 (11) , 1-53
Top Cited PapersOpen AccessLarge Language Models for Combinatorial Optimization: A Systematic Review
This systematic review explores the application of Large Language Models (LLMs) in Combinatorial Optimization (CO). We report our findings using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. We conduct a literature search via Scopus and Google Scholar, examining over 2,000 publications. We assess publications against four inclusion and four exclusion criteria related to their language, research focus, publication year, and type. Eventually, we select 103 studies. We classify these studies into semantic categories and topics to provide a comprehensive overview of the field, including the tasks performed by LLMs, the architectures of LLMs, the existing datasets specifically designed for evaluating LLMs in CO, and the field of application. Finally, we identify future directions for leveraging LLMs in this field.
- Open Access
- 28 March 2026
- Published by Elsevier in SSRN Electronic Journal
Revisiting Local Search for Curriculum-Based Course Timetabling
Curriculum-Based Course Timetabling (CB‐CTT) is a well‐established benchmark problem in educational timetabling, with numerous real and synthetic datasets. Despite its long history of algorithmic developments, designing search methods that work effectively across diverse datasets under time limits remains challenging. In this paper, we focus on local search and, after a thorough analysis of all neighborhood structures reported in the literature, we propose an improved Simulated Annealing (SA) algorithm driven by an extended version of the classic LectureMove neighborhood, called LectureKick, that allows flexible lecture relocation while preserving feasibility. The neighborhood is further enriched with several sampling biases specifically designed to address the soft constraints of CB-CTT, while the overall behavior of the SA is regulated by a cut-off mechanism embedded in the cooling schedule.The method involves numerous parameters, both for the SA schedule and for the neighborhood biases, which are rigorously tuned on a large artificial dataset. The final algorithm is subsequently evaluated on the available benchmarks, including the "standard" ITC~2007 comp dataset, under the original competition timeout conditions.The results show that our approach clearly outperforms all previously proposed local search methods and is competitive with the best methods reported for CB-CTT overall, while improving upon them on several instances.
- 28 March 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessRevisiting Local Search for Curriculum-Based Course Timetabling
Curriculum-Based Course Timetabling (CB‐CTT) is a well‐established benchmark problem in educational timetabling, with numerous real and synthetic datasets. Despite its long history of algorithmic developments, designing search methods that work effectively across diverse datasets under time limits remains challenging. In this paper, we focus on local search and, after a thorough analysis of all neighborhood structures reported in the literature, we propose an improved Simulated Annealing (SA) algorithm driven by an extended version of the classic LectureMove neighborhood, called LectureKick, that allows flexible lecture relocation while preserving feasibility. The neighborhood is further enriched with several sampling biases specifically designed to address the soft constraints of CB-CTT, while the overall behavior of the SA is regulated by a cut-off mechanism embedded in the cooling schedule.The method involves numerous parameters, both for the SA schedule and for the neighborhood biases, which are rigorously tuned on a large artificial dataset. The final algorithm is subsequently evaluated on the available benchmarks, including the "standard" ITC~2007 comp dataset, under the original competition timeout conditions.The results show that our approach clearly outperforms all previously proposed local search methods and is competitive with the best methods reported for CB-CTT overall, while improving upon them on several instances.
- Open Access
Assessing Reproducibility in Evolutionary Computation: A Case Study using Human- and LLM-based Assessment
Reproducibility is an important requirement in evolutionary computation, where results largely depend on computational experiments. In practice, reproducibility relies on how algorithms, experimental protocols, and artifacts are documented and shared. Despite growing awareness, there is still limited empirical evidence on the actual reproducibility levels of published work in the field. In this paper, we study the reproducibility practices in papers published in the Evolutionary Combinatorial Optimization and Metaheuristics track of the Genetic and Evolutionary Computation Conference over a ten-year period. We introduce a structured reproducibility checklist and apply it through a systematic manual assessment of the selected corpus. In addition, we propose RECAP (REproducibility Checklist Automation Pipeline), an LLM-based system that automatically evaluates reproducibility signals from paper text and associated code repositories. Our analysis shows that papers achieve an average completeness score of 0.62, and that 36.90% of them provide additional material beyond the manuscript itself. We demonstrate that automated assessment is feasible: RECAP achieves substantial agreement with human evaluators (Cohen's k of 0.67). Together, these results highlight persistent gaps in reproducibility reporting and suggest that automated tools can effectively support large-scale, systematic monitoring of reproducibility practices.
Open AccessAssessing Reproducibility in Evolutionary Computation: A Case Study using Human- and LLM-based Assessment
Reproducibility is an important requirement in evolutionary computation, where results largely depend on computational experiments. In practice, reproducibility relies on how algorithms, experimental protocols, and artifacts are documented and shared. Despite growing awareness, there is still limited empirical evidence on the actual reproducibility levels of published work in the field. In this paper, we study the reproducibility practices in papers published in the Evolutionary Combinatorial Optimization and Metaheuristics track of the Genetic and Evolutionary Computation Conference over a ten-year period. We introduce a structured reproducibility checklist and apply it through a systematic manual assessment of the selected corpus. In addition, we propose RECAP (REproducibility Checklist Automation Pipeline), an LLM-based system that automatically evaluates reproducibility signals from paper text and associated code repositories. Our analysis shows that papers achieve an average completeness score of 0.62, and that 36.90% of them provide additional material beyond the manuscript itself. We demonstrate that automated assessment is feasible: RECAP achieves substantial agreement with human evaluators (Cohen's k of 0.67). Together, these results highlight persistent gaps in reproducibility reporting and suggest that automated tools can effectively support large-scale, systematic monitoring of reproducibility practices.
- Open Access
- 30 December 2025
- Published by Wiley in International Transactions in Operational Research
A unified formulation for home healthcare routing and scheduling problems
Home Healthcare is an essential component of healthcare systems, where caregivers visit patients' homes to deliver services. While presenting advantages with respect to institutional care, such as being cost‐effective and alleviating family burdens, it presents challenges in scheduling and routing caregivers efficiently. While various formulations of such a problem—the home healthcare routing and scheduling problem—have been proposed, many fail to include key real‐world features, limiting their applicability in practice. This paper consolidates relevant features from existing models and significantly extends the formulation by Mankowska et al. (2014) to develop a unified framework incorporating as many real‐world aspects as possible. We introduce a new instance generator, a set of real‐world features to describe these instances, and a comprehensive infrastructure for validating and comparing solutions. Furthermore, we extend two state‐of‐the‐art solution methods—simulated annealing and one based on a compact mixed integer linear programming model solved by a state‐of‐the‐art software – to integrate all these features and provide a comparative analysis of their performance. Results show that our general solution methods are able to outperform methods specialized for a specific formulation on many instances.
- 30 December 2025
- Published by Wiley in International Transactions in Operational Research
Open AccessA unified formulation for home healthcare routing and scheduling problems
Home Healthcare is an essential component of healthcare systems, where caregivers visit patients' homes to deliver services. While presenting advantages with respect to institutional care, such as being cost‐effective and alleviating family burdens, it presents challenges in scheduling and routing caregivers efficiently. While various formulations of such a problem—the home healthcare routing and scheduling problem—have been proposed, many fail to include key real‐world features, limiting their applicability in practice. This paper consolidates relevant features from existing models and significantly extends the formulation by Mankowska et al. (2014) to develop a unified framework incorporating as many real‐world aspects as possible. We introduce a new instance generator, a set of real‐world features to describe these instances, and a comprehensive infrastructure for validating and comparing solutions. Furthermore, we extend two state‐of‐the‐art solution methods—simulated annealing and one based on a compact mixed integer linear programming model solved by a state‐of‐the‐art software – to integrate all these features and provide a comparative analysis of their performance. Results show that our general solution methods are able to outperform methods specialized for a specific formulation on many instances.
- Open Access
Behavior and Representation in Large Language Models for Combinatorial Optimization: From Feature Extraction to Algorithm Selection
Recent advances in Large Language Models (LLMs) have opened new perspectives for automation in optimization. While several studies have explored how LLMs can generate or solve optimization models, far less is understood about what these models actually learn regarding problem structure or algorithmic behavior. This study investigates how LLMs internally represent combinatorial optimization problems and whether such representations can support downstream decision tasks. We adopt a twofold methodology combining direct querying, which assesses LLM capacity to explicitly extract instance features, with probing analyses that examine whether such information is implicitly encoded within their hidden layers. The probing framework is further extended to a per-instance algorithm selection task, evaluating whether LLM-derived representations can predict the best-performing solver. Experiments span four benchmark problems and three instance representations. Results show that LLMs exhibit moderate ability to recover feature information from problem instances, either through direct querying or probing. Notably, the predictive power of LLM hidden-layer representations proves comparable to that achieved through traditional feature extraction, suggesting that LLMs capture meaningful structural information relevant to optimization performance.
Open AccessBehavior and Representation in Large Language Models for Combinatorial Optimization: From Feature Extraction to Algorithm Selection
Recent advances in Large Language Models (LLMs) have opened new perspectives for automation in optimization. While several studies have explored how LLMs can generate or solve optimization models, far less is understood about what these models actually learn regarding problem structure or algorithmic behavior. This study investigates how LLMs internally represent combinatorial optimization problems and whether such representations can support downstream decision tasks. We adopt a twofold methodology combining direct querying, which assesses LLM capacity to explicitly extract instance features, with probing analyses that examine whether such information is implicitly encoded within their hidden layers. The probing framework is further extended to a per-instance algorithm selection task, evaluating whether LLM-derived representations can predict the best-performing solver. Experiments span four benchmark problems and three instance representations. Results show that LLMs exhibit moderate ability to recover feature information from problem instances, either through direct querying or probing. Notably, the predictive power of LLM hidden-layer representations proves comparable to that achieved through traditional feature extraction, suggesting that LLMs capture meaningful structural information relevant to optimization performance.
- Open Access
Large Language Models for Combinatorial Optimization: A Systematic Review
This systematic review explores the application of Large Language Models (LLMs) in Combinatorial Optimization (CO). We report our findings using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. We conduct a literature search via Scopus and Google Scholar, examining over 2,000 publications. We assess publications against four inclusion and four exclusion criteria related to their language, research focus, publication year, and type. Eventually, we select 103 studies. We classify these studies into semantic categories and topics to provide a comprehensive overview of the field, including the tasks performed by LLMs, the architectures of LLMs, the existing datasets specifically designed for evaluating LLMs in CO, and the field of application. Finally, we identify future directions for leveraging LLMs in this field.
Open AccessLarge Language Models for Combinatorial Optimization: A Systematic Review
This systematic review explores the application of Large Language Models (LLMs) in Combinatorial Optimization (CO). We report our findings using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. We conduct a literature search via Scopus and Google Scholar, examining over 2,000 publications. We assess publications against four inclusion and four exclusion criteria related to their language, research focus, publication year, and type. Eventually, we select 103 studies. We classify these studies into semantic categories and topics to provide a comprehensive overview of the field, including the tasks performed by LLMs, the architectures of LLMs, the existing datasets specifically designed for evaluating LLMs in CO, and the field of application. Finally, we identify future directions for leveraging LLMs in this field.