Using pre-trained large language models (LLMs) as a backbone for time series prediction has recently attracted growing research interest. Existing approaches typically split time series into patches, map them to the token space of LLMs via a Tokenizer, process the tokens through a frozen or fine-tuned LLM backbone, and then reconstruct numerical forecasts using a Detokenizer. However, the actual effectiveness of LLMs for time series forecasting remains under debate. We observe that when trained and evaluated on small datasets, the Tokenizer–Detokenizer components often overfit to the specific data distribution, thereby masking the intrinsic predictive capability of the LLM backbone. To investigate the inherent potential of LLMs in this context, we design three models with identical architectures but distinct pre-training strategies. By leveraging large-scale pre-training, we obtain more unbiased Tokenizer–Detokenizer pairs that are seamlessly integrated with the LLM backbone. Through controlled experiments, we evaluate the zero-shot and few-shot forecasting performance of the LLM, offering insights into its true capabilities. Our extensive experiments reveal that, although the LLM backbone shows some promise, its performance remains limited and does not consistently surpass that of models specifically trained on large-scale time series data. Our source code is publicly available in the repository: https://github.com/SiriZhang45/LLM4TS.

As user experience (UX) increasingly shapes the value and usability of digital products, enabling non-designers to transform their ideas into coherent UX outcomes remains an important yet challenging problem. Despite advances in AI-assisted tools, UX design still requires cross-stage reasoning across strategic, structural, and interface layers, which is often unsupported by existing systems. To address this gap, this study proposes MAUX, a layered multi-agent UX framework that structures and coordinates design reasoning through goal-driven workflows. MAUX operationalizes the five classical UX layers (strategy, scope, structure, skeleton, and surface) via LLM-driven agents grounded in human–computer interaction principles. A blackboard architecture coordinated by a Meta-Agent maintains semantic consistency across stages, enabling systematic progression from initial intent to interface realization. The framework is demonstrated through the design of Urban AirLink, a conceptual low-altitude mobility platform, and evaluated against four LLM-based baselines in a controlled user study ( N = 50 ). Results show that MAUX significantly improves design quality, structural clarity, and cross-layer goal alignment compared to baseline approaches. These findings highlight the potential of MAUX to support reasoning-centered and democratized AI-assisted UX design.

Long-term time series forecasting (LTSF) aims to predict future values for a long time based on historical data. The period term is an essential component of the time series, which is complex yet important for LTSF. Although existing studies have achieved promising results, they still have limitations in modeling dynamic complicated periods. Most studies only focus on static periods with fixed time steps, while very few studies attempt to capture dynamic periods in the time domain. In this paper, we dissect the original time series in time and frequency domains and empirically find that changes in periods are more easily captured and quantified in the frequency domain. Based on this observation, we propose to explore dynamic period features using rotation in the frequency domain. To this end, we develop the frequency-based rotation network (FRNet), a novel LTSF method to effectively capture the features of the dynamic complicated periods. FRNet decomposes the original time series into period and trend components. Based on the complex-valued linear networks, it leverages a period frequency rotation module to predict the period component and a patch frequency rotation module to predict the trend component, respectively. Extensive experiments on seven real-world datasets consistently demonstrate the superiority of FRNet over various state-of-the-art methods. The source code is available at https://github.com/SiriZhang45/FRNet.

Recently, network embedding has attracted extensive research interest. Most existing network embedding models are based on Euclidean spaces. However, Euclidean embedding models cannot effectively capture complex patterns, especially latent hierarchical structures underlying in real-world graphs. Consequently, hyperbolic representation models have been developed to preserve the hierarchical information. Nevertheless, existing hyperbolic models only capture the first-order proximity between nodes. To this end, we propose a new embedding model, named Node2LV, that learns the hyperbolic representations of nodes using squared Lorentzian distances. This yields three advantages. First, our model can effectively capture hierarchical structures that come from the network topology. Second, compared with the conventional hyperbolic embedding methods that use computationally expensive Riemannian gradients, it can be optimized in a more efficient way. Lastly, different from existing hyperbolic embedding models, Node2LV captures higher-order proximities. Specifically, we represent each node with two hyperbolic embeddings, and make the embeddings of related nodes close to each other. To preserve higher-order node proximity, we use a random walk strategy to generate local neighborhood context. We conduct extensive experiments on four different types of real-world networks. Empirical results demonstrate that Node2LV significantly outperforms various graph embedding baselines.

With the increasing popularity of location-aware social media services, next-Point-of-Interest (POI) recommendation has gained significant research interest. The key challenge of next-POI recommendation is to precisely learn users' sequential movements from sparse check-in data. To this end, various embedding methods have been proposed to learn the representations of check-in data in the Euclidean space. However, their ability to learn complex patterns, especially hierarchical structures, is limited by the dimensionality of the Euclidean space. To this end, we propose a new research direction that aims to learn the representations of check-in activities in a hyperbolic space, which yields two advantages. First, it can effectively capture the underlying hierarchical structures, which are implied by the power-law distributions of user movements. Second, it provides high representative strength and enables the check-in data to be effectively represented in a low-dimensional space. Specifically, to solve the next-POI recommendation task, we propose a novel hyperbolic metric embedding (HME) model, which projects the check-in data into a hyperbolic space. The HME jointly captures sequential transition, user preference, category and region information in a unified approach by learning embeddings in a shared hyperbolic space. To the best of our knowledge, this is the first study to explore a non-Euclidean embedding model for next-POI recommendation. We conduct extensive experiments on three check-in datasets to demonstrate the superiority of our hyperbolic embedding approach over the state-of-the-art next-POI recommendation algorithms. Moreover, we conduct experiments on another four online transaction datasets for next-item recommendation to further demonstrate the generality of our proposed model.