Recent advancements in large language models (LLMs) have enabled in-context learning (ICL)-based methods that significantly outperform fine-tuning approaches for text-to-SQL tasks. However, their performance is still considerably lower than that of human experts on benchmarks that include complex schemas and queries, such as BIRD. This study considers the sensitivity of LLMs to the prompts and introduces a novel approach that leverages multiple prompts to explore a broader search space for possible answers and effectively aggregate them. Specifically, we robustly refine the database schema through schema linking using multiple prompts. Thereafter, we generate various candidate SQL queries based on the refined schema and diverse prompts. Finally, the candidate queries are filtered based on their confidence scores, and the optimal query is obtained through a multiple-choice selection that is presented to the LLM. When evaluated on the BIRD and Spider benchmarks, the proposed method achieved execution accuracies of 65.5% and 89.6%, respectively, significantly outperforming previous ICL-based methods.

News recommendation systems traditionally rely on extensive real user interaction data to personalize content, which is often inaccessible and raises privacy concerns, particularly in regions lacking such data. To address these challenges, we propose LAUS (LLM As User Simulator), a novel framework that leverages LLM to simulate user interactions for training a news recommender without real user data. Our framework consists of two stages. First, we simulate user clicks by using an LLM to generate click probabilities for candidate news articles based on simulated user reading histories, with candidate sets generated through random sampling and embedding-based retrieval. Second, we train a lightweight news recommender using these simulated user interactions. Experiments on three datasets demonstrate that a news recommender trained with simulated data outperforms models using LLM prompting, while significantly reducing recommendation time per user. Our findings highlight the potential of using LLMs as user simulators to address data scarcity in news recommendation systems.

Query-Focused Summarization (QFS) is a task that aims to extract essential information from a long document and organize it into a summary that can answer a query. Recently, Transformer-based summarization models have been widely used in QFS. However, the simple Transformer architecture cannot utilize the relationships between distant words and information from a query directly. In this study, we propose the QSG Transformer, a novel QFS model that leverages structure information on Query-attentive Semantic Graph (QSG) to address these issues. Specifically, in the QSG Transformer, QSG node representation is improved by a proposed query-attentive graph attention network, which spreads the information of the query node into QSG using Personalized PageRank, and it is used to generate a summary that better reflects the information from the relationships of a query and document. The proposed method is evaluated on two QFS datasets, and it achieves superior performances over the state-of-the-art models.