Abstract
Advancements in large language model (LLM)-based conversational assistants have transformed search experiences into more natural and context-aware dialogues that resemble human conversation. However, limited access to interaction log data hinders a deeper understanding of their real-world usage. To address this gap, we analyzed 16,952 prompt logs from 904 unique users of Labadain Chat, an LLM-based conversational assistant designed for Tetun speakers, to uncover patterns in user search behavior, engagement, and intent. Our findings show that most users (29.87%) spent between one and five minutes per session, with an average of 43 unique daily users. The majority (93.97%) submitted multiple prompts per session, with an average session duration of 16.9 minutes. Most users (95.22%) were based in Timor-Leste, with education and science (28.75%) and health (28.00%) being the most searched topics. We compared our findings with a study on Google Bard logs in English, revealing similar search characteristics-including engagement duration, command-based instructions, and requests for specific assistance. Furthermore, a comparison with two conventional search engines suggests that LLM-based conversational systems have influenced user search behavior on traditional platforms, reflecting a broader trend toward command-driven queries. These insights contribute to a deeper understanding of how user search behavior evolves, particularly within low-resource language communities. To support future research, we publicly release LabadainLog-17k+, a dataset of over 17,000 real-world user search logs in Tetun, offering a unique resource for investigating conversational search in this language.

Abstract
A comparative analysis of Ka and D-band unit cells is presented using a Waveguide Simulator and infinite array models with a Floquet port. Initially, a single-unit cell design is employed with a tapered transition section. Subsequently, a 1 x 2-unit cell is designed and integrated into standard rectangular waveguides WR-34 and WR-7. For the Ka-band, the results obtained from both models exhibit excellent agreement in terms of magnitude and phase. In the D-band, the 1 x 2-unit cell demonstrated low loss for both techniques, and the phase responses were reasonably accurate with differences of less than 40 degrees. At such high frequencies (145-175 GHz), the Waveguide Simulator offers a viable solution for assessing the behavior of the unit cell without the need for a full array.

Abstract
With the recent advancements in the development of hybrid offshore parks and the expected large-scale implementation of them in the near future, it becomes paramount to investigate proper energy management strategies to improve the integrability of these parks into the power systems. This paper addresses a multiobjective energy management approach using a hybrid energy storage system comprising batteries and hydrogen/fuel-cell systems applied to multi-source wind-wave and wind-solar offshore parks to maximize the delivered energy while minimizing the variations of the power output. To find the solution of the optimization problem defined for energy management, a strategy is proposed based on the examination of a set of weighting factors to form the Pareto front while the problem associated with each of them is assessed in a mixed-integer linear programming framework. Subsequently, fuzzy decision making is applied to select the final solution among the ones existing in the Pareto front. The studies are implemented in different locations considering scenarios for electrical system limitation and the place of the storage units. According to the results, applying the proposed multiobjective framework successfully addresses the enhancement of energy delivery and the decrease in power output fluctuations in the hybrid offshore parks across all scenarios of electrical system limitation and combinational storage locations. Based on the results, in addition to the increase in delivered energy, a decrease in power variations by around 40 % up to over 80 % is observed in the studied cases.

Abstract
The growing deployment of artificial intelligence in critical domains exposes a pressing challenge: how reliably models make predictions for ambiguous data without exhibiting overconfidence. We introduce hubris benchmarking, a methodology to evaluate overconfidence in machine learning models. The benchmark is based on a novel architecture, ambiguous generative adversarial networks (AmbiGANs), which are trained to synthesize realistic yet ambiguous datasets. We also propose the hubris metric to quantitatively measure the extent of model overconfidence when faced with these ambiguous images. We illustrate the usage of the methodology by estimating the hubris of state-of-the-art pre-trained models (ConvNext and ViT) on binarized versions of public datasets, including MNIST, Fashion-MNIST, and Pneumonia Chest X-ray. We found that, while ConvNext is on average 3% more accurate than ViT, it often makes excessively confident predictions, on average by 10% points higher than ViT. These results illustrate the usefulness of hubris benchmarking in high-stakes decision processes. © 2025 Elsevier B.V., All rights reserved.

Abstract
Rebeca is 20+ years old. Introduced by Marjan Sirjani and colleagues for modelling and analysing actor-based systems, it comes with a variety of tool support, including dedicated model checkers, simulators, and code generators. When encountering Rebeca for the first time, either as a student, as a researcher, or as a practitioner from industry, one needs to grasp the subtleties of Rebeca ’s semantics, which includes variants with probabilities and time. This paper presents a user-friendly web-based front-end, based on the Caos library for Scala, to animate different operational semantics of (timed) Rebeca. This can facilitate the dissemination and awareness of Rebeca, provide insights into the differences among existing semantics, and support quick experimentation of new variants (e.g., when the order of received messages is preserved). The tool is illustrated by means of a ticket service use case from the literature. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Abstract
Local flexibility markets are a promising solution to aid system operators in managing the network as it faces the growth of distributed resources and the resulting impacts on voltage control, among other factors. This paper presents and simulates a proposal for an intra-day local flexibility market based on grid segmentation. The design provides a market-based solution for distribution system operators (DSOs) to address near-real-time grid issues. The grid segmentation computes the virtual buses that represent each zone and the sensitivity indices that approximate the impact of activating active power flexibility in the buses within the zone. This approach allows DSOs to manage and publish their flexibility needs per zone and enables aggregators to offer flexibility by optimizing their resource portfolios per zone. The simulation outcomes allow for the assessment of market performance according to the number of zones computed and show that addressing overloading and voltage control through zonal approaches can be cost-effective and counterbalance minor errors compared to node-based approaches.