不要停止多方对话！通过约束生成合成的书面多方对话
Don't Stop the Multi-Party! On Generating Synthetic Written Multi-Party Conversations with Constraints
AI总结：
本文探讨利用指令微调的大语言模型生成合成书面多方对话，通过确定性约束如对话结构和参与者立场，发现逐轮生成比一次性生成更符合约束且语言多样性更高。
摘要：
书面多方对话（WMPCs）在跨学科中被广泛研究，社交媒体因其可及性为主要数据来源。然而，这些数据集引发隐私问题且常反映平台特定属性。例如，说话者之间的互动可能受限于僵化的平台结构（如线程、树状讨论），导致过于简单的互动模式（如一对一的“回复”链接）。本工作探讨通过提供确定性约束如对话结构和参与者立场，利用指令微调的大语言模型（LLMs）生成合成WMPCs的可行性。我们研究了两种互补的LLM利用策略：（i）LLM作为WMPC生成器，任务LLM一次性生成完整的WMPC；（ii）LLM作为WMPC参与者，LLM逐轮生成对话（由说话者、收件人和信息组成），并提供对话历史。我们接下来引入分析框架来评估两种策略对约束、内容质量和互动复杂性的符合程度。最后，我们通过人工和LLM作为评判者评估所获得的WMPCs。我们发现LLMs之间存在显著差异，只有部分模型能够生成高质量的WMPCs。我们还发现逐轮生成比一次性生成更符合约束且语言多样性更高。尽管如此，我们的结构和定性评估表明，两种生成策略都能产生高质量的WMPCs。
英文摘要：
Written Multi-Party Conversations (WMPCs) are widely studied across disciplines, with social media as a primary data source due to their accessibility. However, these datasets raise privacy concerns and often reflect platform-specific properties. For example, interactions between speakers may be limited due to rigid platform structures (e.g., threads, tree-like discussions), which yield overly simplistic interaction patterns (e.g., one-to-one "reply-to" links). This work explores the feasibility of generating synthetic WMPCs with instruction-tuned Large Language Models (LLMs) by providing deterministic constraints such as dialogue structure and participants' stance. We investigate two complementary strategies of leveraging LLMs in this context: (i.) LLMs as WMPC generators, where we task the LLM to generate a whole WMPC at once and (ii.) LLMs as WMPC parties, where the LLM generates one turn of the conversation at a time (made of speaker, addressee and message), provided the conversation history. We next introduce an analytical framework to evaluate compliance with the constraints, content quality, and interaction complexity for both strategies. Finally, we assess the level of obtained WMPCs via human and LLM-as-a-judge evaluations. We find stark differences among LLMs, with only some being able to generate high-quality WMPCs. We also find that turn-by-turn generation yields better conformance to constraints and higher linguistic variability than generating WMPCs in one pass. Nonetheless, our structural and qualitative evaluation indicates that both generation strategies can yield high-quality WMPCs.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
