Graph-Based Self-Adaptive Conversational Agent With Context-Awareness Behaviour Predictions

• Main Author: Zhang, Lan (Author)
• Other Authors: Li, Weihua (Contributor)
• Format: Others
• Published: Auckland University of Technology, 2021-10-26T00:31:12Z.
• Subjects: Self-adaptive; Conversational agents; Knowledge graph; Contextual information extraction; Context-awareness; Recommendation system; User behaviour analysis; Thesis
• Online Access: Get fulltext

 Conversational agents have been widely adopted in dialogue systems for various business purposes. Many existing conversational agents are rule-based and require significant human intervention to adapt the knowledge and conversational flow. In this thesis, I propose a graph-based adaptive conversational agent model which is capable of learning knowledge from human beings and adapting the knowledge-base according to user-agent interactions. Extensive experiments have been conducted to evaluate the proposed model by comparing the responses from the proposed adaptive agent model and a conventional agent. The user's personalised knowledge graph is generated through user-agent interactions. To initiate conversations, it is important to develop a model for learning the users' preferences and giving reasonable suggestions during the conversations. Therefore, based on the personalised knowledge collected by the conversational agent, I propose a novel graph-based context-aware user behaviour prediction and recommendation system. The user's preferences have been considered as an important factor in determining the recommendation outputs. On top of that, the model incorporates contextual information extracted from both users' historical behaviours and events relations, where the contexts have been modelled as knowledge graphs. By leveraging the advantages offered from the knowledge graph, events dependencies and their subtle relations can be established and have been introduced into the recommendation process. Experimental results explicitly indicate that the proposed approach can outperform state-of-the-art algorithms and yield better recommendation outcomes.