What You Say in the Conversation Affects the Flow: Building a Model for Conversational Flow Using NLP Methods

你在对话中所说的话会影响流程：使用 NLP 方法构建对话流程模型

• 批准号: 2887095
• 金额: --
• 依托单位: Imperial College London
• 依托单位国家: 英国
• 项目类别: Studentship
• 财政年份: 2023
• 资助国家: 英国
• 起止时间: 2023 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=studentship-2887095
• 关键词: What; Say; Conversation; Affects; Flow

"No Man Is an Island" - John Donne (1624)Humans are inherently social beings; we interact with each other constantly. Conversations, one of the most prevalent social behaviors, play crucial roles in our lives. They allow us to connect, communicate, build relationships, and achieve our goals. In organizations, conversations are fundamental for effective collaboration and decision-making. Successful conversations build trust, foster positive relationships among team members, improve team dynamics, and lead to better problem-solving and innovation, ultimately driving better performance towards collective goals. However, not all conversations are easy to navigate. Various challenges arise when individuals interact and work together, potentially impacting workplace relationships and performance negatively. For instance, initiating conversations with strangers at networking events can be daunting. In teams, conflicts may occur due to differences in personality, communication styles, or goals, and individuals may make mistakes that affect their team's performance or relationships. Previous research on conversations has been limited due to the difficulty of capturing and analyzing natural conversations.In my research, I aim to combine cutting-edge methods such as natural language processing (NLP), machine learning, and experiments to uncover the underlying mechanisms of human conversations and develop practical interventions to help individuals navigate important conversations in everyday life and organizational settings.My first paper "What You Say in the Conversation Affects the Flow: Modelling Conversational Flow Using NLP Methods" has been accepted for academic presentations at the Royal Society scientific meeting, the International Association for Relationship Research annual conference, and the Towards the Future of AI conference in 2022. In this study, I build on recent research on conversational flow and use advanced computational methods to investigate how the content of a conversation influences its flow. I analyze real-world networking conversation data and develop practical interventions to improve conversational flow in networking.In my Second Paper "Understanding Self-Focus and Other-Focus in Conversations", I investigate the fundamental dimension of communication: self-focus versus other-focus. In conversations, we sometimes focus more on ourselves (e.g., redirecting topics to ourselves, bragging) or on our conversation partner (e.g., asking questions, making effective responses). These focuses can be reflected in our words and lead to substantial intrapersonal and interpersonal consequences. Previous research has linked conversational narcissism to lower social attraction (Vangelisti et al., 1990) and found that first-person singular pronoun use relates to speakers' negative emotionality (Berry-Blunt et al., 2021). Building on these works and leveraging advanced computational methods, I aim to identify diverse linguistic features and create a comprehensive understanding of self-focus and other-focus tendencies in conversations.As part of this project, I will visit Harvard Business School to collaborate with my collaborators and experts in this research field, collect data from real conversations in the behavioral lab, and further develop the research. This collaboration is crucial for data collection, analysis, and paper development. Therefore, I will apply for OSFW, as the visit is essential for the advancement and completion of my research projects.In summary, my research leverages cutting edge methods to understand human conversations by unpacking decision-making processes in difficult social situations and develop interventions to improve interactions and relationships in everyday life and organizational contexts.

“没有人是一座孤岛”-约翰·多恩（1624）人类天生是社会性的存在;我们不断地相互作用。会话是最普遍的社会行为之一，在我们的生活中起着至关重要的作用。它们使我们能够连接，沟通，建立关系，并实现我们的目标。在组织中，对话是有效协作和决策的基础。成功的对话建立信任，培养团队成员之间的积极关系，提高团队活力，并导致更好地解决问题和创新，最终推动更好的绩效实现集体目标。然而，并不是所有的对话都很容易导航。当个人互动和一起工作时，会出现各种挑战，可能会对工作场所的关系和绩效产生负面影响。例如，在社交活动中与陌生人开始对话可能会令人望而生畏。在团队中，由于个性、沟通方式或目标的差异，可能会发生冲突，个人可能会犯错误，影响团队的表现或关系。由于难以捕捉和分析自然对话，以前对对话的研究一直受到限制。在我的研究中，我的目标是联合收割机自然语言处理（NLP），机器学习，和实验，以揭示人类对话的潜在机制，并制定切实可行的干预措施，以帮助个人导航在日常生活和组织设置的重要对话。我的第一篇论文“你在对话中说的话会影响交流：使用NLP方法建模对话交流”已被皇家学会科学会议、国际关系研究协会年会和2022年人工智能未来会议的学术报告所接受。在这项研究中，我建立在最近的研究会话流，并使用先进的计算方法来调查如何会话的内容影响其流量。我分析了现实世界的网络会话数据，并制定了切实可行的干预措施，以改善会话流在network.In我的第二篇论文“理解自我的焦点，并在会话焦点”，我调查了沟通的基本维度：自我焦点与他人焦点。在谈话中，我们有时会更关注自己（例如，将话题重定向到我们自己、吹牛）或我们的谈话伙伴（例如，提出问题，作出有效的回应）。这些焦点可以反映在我们的话语中，并导致实质性的内在和人际后果。先前的研究已经将会话自恋与较低的社会吸引力联系起来（Vangelisti等人，1990），并发现第一人称单数代词的使用与说话者的负面情绪有关（Berry-Blunt et al.，2021年）。在此基础上，利用先进的计算方法，我希望能够识别出不同的语言特征，并对对话中的自我关注和他人关注倾向有一个全面的了解。作为这个项目的一部分，我将访问哈佛商学院，与我的合作者和这一研究领域的专家合作，在行为实验室收集真实的对话数据，并进一步发展研究。这种合作对于数据收集、分析和论文开发至关重要。因此，我将申请OSFW，因为这次访问对我的研究项目的推进和完成至关重要。总而言之，我的研究利用尖端方法，通过在困难的社会情况下拆开决策过程来理解人类对话，并制定干预措施，以改善日常生活和组织环境中的互动和关系。