Showing 1–6 of 6 results for author: De Lima, V

Passage Query Methods for Retrieval and Reranking in Conversational Agents

Authors: Victor De Lima, Grace Hui Yang

Abstract: This paper presents our approach to the TREC Interactive Knowledge Assistance Track (iKAT), which focuses on improving conversational information-seeking (CIS) systems. While recent advancements in CIS have improved conversational agents' ability to assist users, significant challenges remain in understanding context and retrieving relevant documents across domains and dialogue turns. To address t… ▽ More This paper presents our approach to the TREC Interactive Knowledge Assistance Track (iKAT), which focuses on improving conversational information-seeking (CIS) systems. While recent advancements in CIS have improved conversational agents' ability to assist users, significant challenges remain in understanding context and retrieving relevant documents across domains and dialogue turns. To address these issues, we extend the Generate-Retrieve-Generate pipeline by developing passage queries (PQs) that align with the target document's expected format to improve query-document matching during retrieval. We propose two variations of this approach: Weighted Reranking and Short and Long Passages. Each method leverages a Meta Llama model for context understanding and generating queries and responses. Passage ranking evaluation results show that the Short and Long Passages approach outperformed the organizers' baselines, performed best among Llama-based systems in the track, and achieved results comparable to GPT-4-based systems. These results indicate that the method effectively balances efficiency and performance. Findings suggest that PQs improve semantic alignment with target documents and demonstrate their potential to improve multi-turn dialogue systems. △ Less

Submitted 28 February, 2025; originally announced March 2025.

Comments: 7 pages, 3 figures. In Proceedings of the Thirty-Third Text Retrieval Conference (TREC 2024), November 18-22, 2024, Rockville, MD, USA

Algorithms for the Bin Packing Problem with Scenarios

Authors: Yulle G. F. Borges, Vinícius L. de Lima, Flávio K. Miyazawa, Lehilton L. C. Pedrosa, Thiago A. de Queiroz, Rafael C. S. Schouery

Abstract: This paper presents theoretical and practical results for the bin packing problem with scenarios, a generalization of the classical bin packing problem which considers the presence of uncertain scenarios, of which only one is realized. For this problem, we propose an absolute approximation algorithm whose ratio is bounded by the square root of the number of scenarios times the approximation ratio… ▽ More This paper presents theoretical and practical results for the bin packing problem with scenarios, a generalization of the classical bin packing problem which considers the presence of uncertain scenarios, of which only one is realized. For this problem, we propose an absolute approximation algorithm whose ratio is bounded by the square root of the number of scenarios times the approximation ratio for an algorithm for the vector bin packing problem. We also show how an asymptotic polynomial-time approximation scheme is derived when the number of scenarios is constant. As a practical study of the problem, we present a branch-and-price algorithm to solve an exponential model and a variable neighborhood search heuristic. To speed up the convergence of the exact algorithm, we also consider lower bounds based on dual feasible functions. Results of these algorithms show the competence of the branch-and-price in obtaining optimal solutions for about 59% of the instances considered, while the combined heuristic and branch-and-price optimally solved 62% of the instances considered. △ Less

Submitted 24 May, 2023; originally announced May 2023.

On the Linguistic and Computational Requirements for Creating Face-to-Face Multimodal Human-Machine Interaction

Authors: João Ranhel, Cacilda Vilela de Lima

Abstract: In this study, conversations between humans and avatars are linguistically, organizationally, and structurally analyzed, focusing on what is necessary for creating face-to-face multimodal interfaces for machines. We videorecorded thirty-four human-avatar interactions, performed complete linguistic microanalysis on video excerpts, and marked all the occurrences of multimodal actions and events. Sta… ▽ More In this study, conversations between humans and avatars are linguistically, organizationally, and structurally analyzed, focusing on what is necessary for creating face-to-face multimodal interfaces for machines. We videorecorded thirty-four human-avatar interactions, performed complete linguistic microanalysis on video excerpts, and marked all the occurrences of multimodal actions and events. Statistical inferences were applied to data, allowing us to comprehend not only how often multimodal actions occur but also how multimodal events are distributed between the speaker (emitter) and the listener (recipient). We also observed the distribution of multimodal occurrences for each modality. The data show evidence that double-loop feedback is established during a face-to-face conversation. This led us to propose that knowledge from Conversation Analysis (CA), cognitive science, and Theory of Mind (ToM), among others, should be incorporated into the ones used for describing human-machine multimodal interactions. Face-to-face interfaces require an additional control layer to the multimodal fusion layer. This layer has to organize the flow of conversation, integrate the social context into the interaction, as well as make plans concerning 'what' and 'how' to progress on the interaction. This higher level is best understood if we incorporate insights from CA and ToM into the interface system. △ Less

Submitted 24 November, 2022; originally announced November 2022.

Exact solution of network flow models with strong relaxations

Authors: Vinícius L. de Lima, Manuel Iori, Flávio K. Miyazawa

Abstract: We address the solution of Mixed Integer Linear Programming (MILP) models with strong relaxations that are derived from Dantzig-Wolfe decompositions and allow a pseudo-polynomial pricing algorithm. We exploit their network-flow characterization and provide a framework based on column generation, reduced-cost variable-fixing, and a highly asymmetric branching scheme that allows us to take advantage… ▽ More We address the solution of Mixed Integer Linear Programming (MILP) models with strong relaxations that are derived from Dantzig-Wolfe decompositions and allow a pseudo-polynomial pricing algorithm. We exploit their network-flow characterization and provide a framework based on column generation, reduced-cost variable-fixing, and a highly asymmetric branching scheme that allows us to take advantage of the potential of the current MILP solvers. We apply our framework to a variety of cutting and packing problems from the literature. The efficiency of the framework is proved by extensive computational experiments, in which a significant number of open instances could be solved to proven optimality for the first time. △ Less

Submitted 31 May, 2021; originally announced May 2021.

Arc Flow Formulations Based on Dynamic Programming: Theoretical Foundations and Applications

Authors: Vinícius L. de Lima, Cláudio Alves, François Clautiaux, Manuel Iori, José M. Valério de Carvalho

Abstract: Network flow formulations are among the most successful tools to solve optimization problems. Such formulations correspond to determining an optimal flow in a network. One particular class of network flow formulations is the arc flow, where variables represent flows on individual arcs of the network. For $\mathcal{NP}$-hard problems, polynomial-sized arc flow models typically provide weak linear r… ▽ More Network flow formulations are among the most successful tools to solve optimization problems. Such formulations correspond to determining an optimal flow in a network. One particular class of network flow formulations is the arc flow, where variables represent flows on individual arcs of the network. For $\mathcal{NP}$-hard problems, polynomial-sized arc flow models typically provide weak linear relaxations and may have too much symmetry to be efficient in practice. Instead, arc flow models with a pseudo-polynomial size usually provide strong relaxations and are efficient in practice. The interest in pseudo-polynomial arc flow formulations has grown considerably in the last twenty years, in which they have been used to solve many open instances of hard problems. A remarkable advantage of pseudo-polynomial arc flow models is the possibility to solve practical-sized instances directly by a Mixed Integer Linear Programming solver, avoiding the implementation of complex methods based on column generation. In this survey, we present theoretical foundations of pseudo-polynomial arc flow formulations, by showing a relation between their network and Dynamic Programming (DP). This relation allows a better understanding of the strength of these formulations, through a link with models obtained by Dantzig-Wolfe decomposition. The relation with DP also allows a new perspective to relate state-space relaxation methods for DP with arc flow models. We also present a dual point of view to contrast the linear relaxation of arc flow models with that of models based on paths and cycles. To conclude, we review the main solution methods and applications of arc flow models based on DP in several domains such as cutting, packing, scheduling, and routing. △ Less

Submitted 15 April, 2021; v1 submitted 1 October, 2020; originally announced October 2020.

Statistics for Ranking Program Committees and Editorial Boards

Authors: Roberto da Silva, Jose Palazzo de Oliveira, Jose Valdeni de Lima, Viviane Moreira

Abstract: Ranking groups of researchers is important in several contexts and can serve many purposes such as the fair distribution of grants based on the scientist's publication output, concession of research projects, classification of journal editorial boards and many other applications in a social context. In this paper, we propose a method for measuring the performance of groups of researchers. The pr… ▽ More Ranking groups of researchers is important in several contexts and can serve many purposes such as the fair distribution of grants based on the scientist's publication output, concession of research projects, classification of journal editorial boards and many other applications in a social context. In this paper, we propose a method for measuring the performance of groups of researchers. The proposed method is called alpha-index and it is based on two parameters: (i) the homogeneity of the h-indexes of the researchers in the group; and (ii) the h-group, which is an extension of the h-index for groups. Our method integrates the concepts of homogeneity and absolute value of the h-index into a single measure which is appropriate for the evaluation of groups. We report on experiments that assess computer science conferences based on the h-indexes of their program committee members. Our results are similar to a manual classification scheme adopted by a research agency. △ Less

Submitted 4 February, 2010; originally announced February 2010.