Scholarlink

This paper discusses ongoing work in the SALLY research project, which aims to help migrants in Greece integrate using conversational agent technology. The main users of SALLY are Third Country Nationals (TCNs) in Greece, who can interact with the agent to get information on topics like legal processes, available services, and other important matters. To improve the chatbot’s performance, it was important to better understand how users interact with SALLY. To support this goal, we developed a secondary chatbot designed to help SALLY administrators analyze user interactions. This chatbot was trained using real conversations between users and SALLY and is intended for administrator use only. User conversations within SALLY are stored along with key details in knowledge graphs. These details include important parts of the conversations, such as mentioned entities, relationships, and private concepts. The stored information is then used to train the secondary chatbot, allowing administrators to gain useful insights from user interactions.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Knowledge Graph-Based Conversational Agent for Analyzing User Interactions

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Christos Bouas,
  • Maria Papoutsoglou,
  • Alexandros Tassios,
  • Stergios Tegos,
  • Konstantinos Manousaridis,
  • Maria Kaltsa,
  • Thanassis Mavropoulos,
  • Stefanos Vrochidis,
  • Georgios Meditskos

摘要

This paper discusses ongoing work in the SALLY research project, which aims to help migrants in Greece integrate using conversational agent technology. The main users of SALLY are Third Country Nationals (TCNs) in Greece, who can interact with the agent to get information on topics like legal processes, available services, and other important matters. To improve the chatbot’s performance, it was important to better understand how users interact with SALLY. To support this goal, we developed a secondary chatbot designed to help SALLY administrators analyze user interactions. This chatbot was trained using real conversations between users and SALLY and is intended for administrator use only. User conversations within SALLY are stored along with key details in knowledge graphs. These details include important parts of the conversations, such as mentioned entities, relationships, and private concepts. The stored information is then used to train the secondary chatbot, allowing administrators to gain useful insights from user interactions.