Tree-of-Debate: Multi-Persona Debate Trees Elicit Critical Thinking for Scientific Comparative Analysis

With the exponential growth of research facilitated by modern technology and improved accessibility, scientific discoveries have become increasingly fragmented within and across fields. This makes it challenging to assess the significance, novelty, incremental findings, and equivalent ideas between related works, particularly those from different research communities. Large language models (LLMs) have recently demonstrated strong quantitative and qualitative reasoning abilities, and multi-agent LLM debates have shown promise in handling complex reasoning tasks by exploring diverse perspectives and reasoning paths. Inspired by this, we introduce Tree-of-Debate (ToD), a framework which converts scientific papers into LLM personas that debate their respective novelties. To emphasize structured, critical reasoning rather than focusing solely on outcomes, ToD dynamically constructs a debate tree, enabling fine-grained analysis of independent novelty arguments within scholarly articles. Through experiments on scientific literature across various domains, evaluated by expert researchers, we demonstrate that ToD generates informative arguments, effectively contrasts papers, and supports researchers in their literature review.

Unsupervised Episode Detection for Large-Scale News Events

Episodic structures are inherently interpretable and adaptable to evolving large-scale key events. However, state-of-the-art automatic event detection methods overlook event episodes and, therefore, struggle with these crucial characteristics. This paper introduces a novel task, episode detection, aimed at identifying episodes from a news corpus containing key event articles. An episode describes a cohesive cluster of core entities (e.g., "protesters", "police") performing actions at a specific time and location. Furthermore, an episode is a significant part of a larger group of episodes under a particular key event. Automatically detecting episodes is challenging because, unlike key events and atomic actions, we cannot rely on explicit mentions of times and locations to distinguish between episodes or use semantic similarity to merge inconsistent episode co-references. To address these challenges, we introduce EpiMine, an unsupervised episode detection framework that (1) automatically identifies the most salient, key-event-relevant terms and segments, (2) determines candidate episodes in an article based on natural episodic partitions estimated through shifts in discriminative term combinations, and (3) refines and forms final episode clusters using large language model-based reasoning on the candidate episodes. We construct three diverse, real-world event datasets annotated at the episode level. EpiMine outperforms all baselines on these datasets by an average 59.2% increase across all metrics.

Instruct, Not Assist: LLM-based Multi-Turn Planning and Hierarchical Questioning for Socratic Code Debugging

Socratic questioning is an effective teaching strategy, encouraging critical thinking and problem-solving. The conversational capabilities of large language models (LLMs) show great potential for providing scalable, real-time student guidance. However, current LLMs often give away solutions directly, making them ineffective instructors. We tackle this issue in the code debugging domain with TreeInstruct, an Instructor agent guided by a novel state space-based planning algorithm. TreeInstruct asks probing questions to help students independently identify and resolve errors. It estimates a student's conceptual and syntactical knowledge to dynamically construct a question tree based on their responses and current knowledge state, effectively addressing both independent and dependent mistakes concurrently in a multi-turn interaction setting. In addition to using an existing single-bug debugging benchmark, we construct a more challenging multi-bug dataset of 150 coding problems, incorrect solutions, and bug fixes -- all carefully constructed and annotated by experts. Extensive evaluation shows TreeInstruct's state-of-the-art performance on both datasets, proving it to be a more effective instructor than baselines. Furthermore, a real-world case study with five students of varying skill levels further demonstrates TreeInstruct's ability to guide students to debug their code efficiently with minimal turns and highly Socratic questioning.

MEGClass: Text Classification with Extremely Weak Supervision via Mutually-Enhancing Text Granularities

Text classification typically requires a substantial amount of human-annotated data to serve as supervision, which is costly to obtain in dynamic emerging domains. Certain methods seek to address this problem by solely relying on the surface text of class names to serve as extremely weak supervision. However, existing methods fail to account for single-class documents discussing multiple topics. Both topic diversity and vague sentences may introduce noise into the document's underlying representation and consequently the precision of the predicted class. Furthermore, current work focuses on text granularities (documents, sentences, or words) independently, which limits the degree of coarse- or fine-grained context that we can jointly extract from all three to identify significant subtext for classification. In order to address this problem, we propose MEGClass, an extremely weakly-supervised text classification method to exploit Mutually-Enhancing Text Granularities. Specifically, MEGClass constructs class-oriented sentence and class representations based on keywords for performing a sentence-level confidence-weighted label ensemble in order to estimate a document's initial class distribution. This serves as the target distribution for a multi-head attention network with a class-weighted contrastive loss. This network learns contextualized sentence representations and weights to form document representations that reflect its original document and sentence-level topic diversity. Retaining this heterogeneity allows MEGClass to select the most class-indicative documents to serve as iterative feedback for enhancing the class representations. Finally, these top documents are used to fine-tune a pre-trained text classifier. As demonstrated through extensive experiments on six benchmark datasets, MEGClass outperforms other weakly and extremely weakly supervised methods.