Number-focused headline generation is a summarization task requiring both high textual quality and precise numerical accuracy, which poses a unique challenge for Large Language Models (LLMs). Existing studies in the literature focus only on either textual quality or numerical reasoning and thus are inadequate to address this challenge. In this paper, we propose a novel chain-of-thought framework for using rationales comprising key elements of the Topic, Entities, and Numerical reasoning (TEN) in news articles to enhance the capability for LLMs to generate topic-aligned high-quality texts with precise numerical accuracy. Specifically, a teacher LLM is employed to generate TEN rationales as supervision data, which are then used to teach and fine-tune a student LLM. Our approach teaches the student LLM automatic generation of rationales with enhanced capability for numerical reasoning and topic-aligned numerical headline generation. Experiments show that our approach achieves superior performance in both textual quality and numerical accuracy.

Large language models (LLMs) face persistent challenges when handling long-context tasks, most notably the “lost in the middle” issue, where information located in the middle of a long input tends to be underutilized. Some existing methods that reduce input have the risk of discarding key information, while others that extend context windows often lead to attention dispersion. To address these limitations, we propose Tree of Agents (TOA), a multi-agent reasoning framework that segments the input into chunks processed by independent agents. Each agent generates its local cognition, then agents dynamically exchange information for collaborative reasoning along tree-structured paths. TOA enables agents to probe different reasoning orders for multi-perspective understanding, effectively mitigating position bias and reducing hallucinations. To improve processing efficiency, we incorporate prefix-hash caching and adaptive pruning strategies, achieving significant performance improvements with comparable API overhead. Experiments show that TOA, powered by compact LLaMA3.1-8B, significantly outperforms multiple baselines and demonstrates comparable performance to the latest and much larger commercial models, such as Gemini1.5-pro, on various long-context tasks. Code is available at https://github.com/Aireduce952/Tree-of-Agents.

In this paper, we present our system for the SemEval-2024 Task 7, i.e., NumEval subtask 3: Numericial Reasoning. Given a news article and its headline, the numerical reasoning task involves creating a system to compute the intentionally excluded number within the news headline. We propose a fine-tuned GPT-3.5-turbo model, specifically engineered to deduce missing numerals directly from the content of news article. The model is trained with a human-engineered prompt that itegrates the news content and the masked headline, tailoring its accuracy for the designated task. It achieves an accuracy of 0.94 on the test data and secures the second position in the official leaderboard. An examination on the system’s inference results reveals its commendable accuracy in identifying correct numerals when they can be directly “copied” from the articles. However, the error rates increase when it comes to some ambiguous operations such as rounding.

Stance detection is an important task, which aims to classify the attitude of an opinionated text towards a given target. Remarkable success has been achieved when sufficient labeled training data is available. However, annotating sufficient data is labor-intensive, which establishes significant barriers for generalizing the stance classifier to the data with new targets. In this paper, we proposed a Semantic-Emotion Knowledge Transferring (SEKT) model for cross-target stance detection, which uses the external knowledge (semantic and emotion lexicons) as a bridge to enable knowledge transfer across different targets. Specifically, a semantic-emotion heterogeneous graph is constructed from external semantic and emotion lexicons, which is then fed into a graph convolutional network to learn multi-hop semantic connections between words and emotion tags. Then, the learned semantic-emotion graph representation, which serves as prior knowledge bridging the gap between the source and target domains, is fully integrated into the bidirectional long short-term memory (BiLSTM) stance classifier by adding a novel knowledge-aware memory unit to the BiLSTM cell. Extensive experiments on a large real-world dataset demonstrate the superiority of SEKT against the state-of-the-art baseline methods.