Search

As language models grow ever larger, the need for large-scale high-quality text datasets has never been more pressing, especially in multilingual settings. The BigScience workshop, a 1-year international and multidisciplinary initiative, was formed with the goal of researching and training large language models as a values-driven undertaking, putting issues of ethics, harm, and governance in the foreground. This paper documents the data creation and curation efforts undertaken by BigScience to assemble the Responsible Open-science Open-collaboration Text Sources (ROOTS) corpus, a 1.6TB dataset spanning 59 languages that was used to train the 176-billion-parameter BigScience Large Open-science Open-access Multilingual (BLOOM) language model. We further release a large initial subset of the corpus and analyses thereof, and hope to empower large-scale monolingual and multilingual modeling projects with both the data and the processing tools, as well as stimulate research around this large multilingual corpus., Comment: NeurIPS 2022, Datasets and Benchmarks Track

This paper presents a user-centered approach to the design and evaluation of conversational search user interfaces to support the pilot in cockpits. Our findings of a controlled user experiment suggest that user perception of the usefulness of the system in completing the search task and the system's responses to the relevance of the topic are good predictors of search performance. User satisfaction with the system's responses may not be a good predictor of search performance in the context of safety of life scenarios such as cockpit procedures for pilots.

Mission planning for Earth Observation Satellite operators typically implies dynamically altering how requests from different customers are prioritised in order to meet expected deadlines. A request corresponds to a given area of interest to capture on Earth. This exercise is challenging for different reasons. First, satellites are limited by maneuvers and power consumption constraints resulting in a limited surface that can be covered at each orbit. Consequently, many requests are in competition with each other and so all of them cannot be treated at each orbit. When a request priority is boosted, it may incidentally penalise surrounding requests. Second, there are several uncertain factors such as weather (in particular cloud cover) and future incoming requests that can impact the completion progress of the requests. With order books of increasing size and the planned operations of a growing number of satellites in a close future, there is a clear need for a decision support method. In this paper, we investigate the potential of Evolutionary Algorithms (EAs) for this problem and propose several approaches to optimise request priorities based on Local Search and Population-Based Incremental Learning (PBIL). Using a certified algorithm to decode request priorities into satellite actions and a satellite simulator developed by Airbus, we are able to realistically evaluate the potential of these methods and benchmark them against operator baselines. Experiments on several scenarios and order books show that EAs can outperform baselines and significantly improve operations both in terms of delay reduction and successful image capture. While black-box approaches yield significant improvements over the baseline when there are many delayed requests in the order books, introducing domain knowledge is required to handle cases with fewer delayed requests.

International audience; The Zambezi Delta draining the Southern African Plateau and the southern part of the East African Rift is one of a the largest delta of Africa with a long-lasting history starting during Early Cretaceous with more than 12 km of sediments deposited. The Zambezi Delta is therefore a unique archive of the past topographic evolution of southern and eastern Africa and their related deformations, but also of the climate changes, global and regional (consequences of local topographic growths). Understanding this archive supposes to get a high-resolution dating of the sediments. Our two objectives are here (1) to construct an age model of the Zambezi Cenozoic delta using a combination of biostratigraphy, orbital stratigraphy and sequence stratigraphy and (2) to determine the palaeoprecipitation variations of the Zambezi catchment from the Oligocene to present day in a known tectonic framework. The Neogene sequences were dated at high-resolution assuming that the third order sequences are of eustatic origin and record long-term eccentricity cycles. The sequences were correlated in ages on the calculated Earth orbital solutions of Laskar for the time intervals provided by the biostratigraphy (nannofossils, planktonic foraminifers). The palaeoprecipitation record was based on the definition of a humidity index based on pollen analysis and associated botanical associations. The late Oligocene was a quite wet period getting dryer in the uppermost Chattian. The base Tortonian (11 Ma) was a humid period. The Messinian was a dry period with a slight increase of the humidity during the Zanclean and a sharp increase around the Zanclean-Piacenzian boundary. The Zambezi Delta has recorded the uplifts of the Southern African Plateau (around 85 Ma and around 25 Ma) and those of the southward migration of the East African Rift (since 5.5 Ma).

This paper describes a software architecture designed to enable the evaluation of information processing and retrieval systems. The overall objective of our project is to provide an open technical framework for the integration of tools for collection, processing, analysis and communication of open source information1. However, enabling the integration of heterogeneous components does not make sense without a proper way to compare the capabilities of multiple tools.Thus, as part of the architecture the VIRTUOSO platform offers an evaluation framework which allows one to deploy and run evaluation kits for different use-cases.

The problems of comparing search support tool in interactive information retrieval (IIR) and of selecting the right one have always been difficult due to the inherent dependency to users. Using an adapted evaluation protocol, we study in this paper different suggestion approaches. The results show that the performance are changing for different users and also during the search sessions. As a consequence, they also show that the selection of a particular support tool has to use new grounding. In this way, we propose a system that allows to combine independent suggestion mechanisms based on an analysis of user behavior and considering the search session time as a key factor instead of using only static rules.

International audience; Nowadays, the core of the Particle Swarm Optimization (PSO) algorithm has proved to be reliable. However, faced with multi-objective problems, adaptations are needed. Deeper researches must be conducted on its key steps, such as solution set management and guide selection, in order to improve its efficiency in this context. Indeed, numerous parameters and implementation strategies can impact on the optimization performance in a particle swarm optimizer. In this paper, our recent works on those topics are presented. We introduce an $\varepsilon$ dominance variation which enables a finer neighborhood handling in criterion space. Then we propose some ideas concerning the guide selection and memorization for each particle. These methods are compared against a standard MOPSO implementation on benchmark problems and against an evolutionary approach (NSGAII) for a real world problem: SVM classifier optimization (or model selection) for a handwritten digits/outliers discrimination problem.

Dupont Gérard. Les problèmes urbains . In: Études Normandes, livraison 69, n°213, 4e trimestre 1968. Les problèmes urbains. pp. 1-12.