Enhancing Sequential Recommendations through Multi-Perspective Reflections and Iteration

Sequence recommendation (SeqRec) aims to predict the next item a user will interact with by understanding user intentions and leveraging collaborative filtering information. Large language models (LLMs) have shown great promise in recommendation tasks through prompt-based, fixed reflection libraries, and fine-tuning techniques. However, these methods face challenges, including lack of supervision, inability to optimize reflection sources, inflexibility to diverse user needs, and high computational costs. Despite promising results, current studies primarily focus on reflections of users' explicit preferences (e.g., item titles) while neglecting implicit preferences (e.g., brands) and collaborative filtering information. This oversight hinders the capture of preference shifts and dynamic user behaviors. Additionally, existing approaches lack mechanisms for reflection evaluation and iteration, often leading to suboptimal recommendations. To address these issues, we propose the Mixture of REflectors (MoRE) framework, designed to model and learn dynamic user preferences in SeqRec. Specifically, MoRE introduces three reflectors for generating LLM-based reflections on explicit preferences, implicit preferences, and collaborative signals. Each reflector incorporates a self-improving strategy, termed refining-and-iteration, to evaluate and iteratively update reflections. Furthermore, a meta-reflector employs a contextual bandit algorithm to select the most suitable expert and corresponding reflections for each user's recommendation, effectively capturing dynamic preferences. Extensive experiments on three real-world datasets demonstrate that MoRE consistently outperforms state-of-the-art methods, requiring less training time and GPU memory compared to other LLM-based approaches in SeqRec.

Exploring the Nexus of Large Language Models and Legal Systems: A Short Survey

With the advancement of Artificial Intelligence (AI) and Large Language Models (LLMs), there is a profound transformation occurring in the realm of natural language processing tasks within the legal domain. The capabilities of LLMs are increasingly demonstrating unique roles in the legal sector, bringing both distinctive benefits and various challenges. This survey delves into the synergy between LLMs and the legal system, such as their applications in tasks like legal text comprehension, case retrieval, and analysis. Furthermore, this survey highlights key challenges faced by LLMs in the legal domain, including bias, interpretability, and ethical considerations, as well as how researchers are addressing these issues. The survey showcases the latest advancements in fine-tuned legal LLMs tailored for various legal systems, along with legal datasets available for fine-tuning LLMs in various languages. Additionally, it proposes directions for future research and development.

Incorporating Judgment Prediction into Legal Case Retrieval via Law-aware Generative Retrieval

Legal case retrieval and judgment prediction are crucial components in intelligent legal systems. In practice, determining whether two cases share the same charges through legal judgment prediction is essential for establishing their relevance in case retrieval. However, current studies on legal case retrieval merely focus on the semantic similarity between paired cases, ignoring their charge-level consistency. This separation leads to a lack of context and potential inaccuracies in the case retrieval that can undermine trust in the system's decision-making process. Given the guidance role of laws to both tasks and inspired by the success of generative retrieval, in this work, we propose to incorporate judgment prediction into legal case retrieval, achieving a novel law-aware Generative legal case retrieval method called Gear. Specifically, Gear first extracts rationales (key circumstances and key elements) for legal cases according to the definition of charges in laws, ensuring a shared and informative representation for both tasks. Then in accordance with the inherent hierarchy of laws, we construct a law structure constraint tree and assign law-aware semantic identifier(s) to each case based on this tree. These designs enable a unified traversal from the root, through intermediate charge nodes, to case-specific leaf nodes, which respectively correspond to two tasks. Additionally, in the training, we also introduce a revision loss that jointly minimizes the discrepancy between the identifiers of predicted and labeled charges as well as retrieved cases, improving the accuracy and consistency for both tasks. Extensive experiments on two datasets demonstrate that Gear consistently outperforms state-of-the-art methods in legal case retrieval while maintaining competitive judgment prediction performance.