ISSN :2582-9793

GE-Blender: Graph Based Knowledge Enhancement for Blender

Original Research (Published On: 25-Oct-2023 )
GE-Blender: Graph Based Knowledge Enhancement for Blender

Xiaolei Lian and Chonglin Tang

Adv. Artif. Intell. Mach. Learn., 3 (4):1572-1587

Xiaolei Lian : Beijing Zhong Ke Lian Yuan Technologies, SDP Department

Chonglin Tang : Beijing Zhong Ke Lian Yuan Technologies, SDP Department

Download PDF Here

Article History: Received on: 21-Jul-23, Accepted on: 18-Oct-23, Published on: 25-Oct-23

Corresponding Author: Xiaolei Lian

Email: lian19931201@gmail.com

Citation: Xiaolei Lian and Chonglin Tang (2023). GE-Blender: Graph Based Knowledge Enhancement for Blender. Adv. Artif. Intell. Mach. Learn., 3 (4 ):1572-1587


Abstract

    In spite of the considerable accomplishments in open-domain dialogue generation, the presence of unseen entities can substantially impact the task, resulting in a performance decline in model-driven dialogue generation. While previous research has utilized re- trieved knowledge of seen entities to augment model representation, a comprehensive exploration of unseen entities, including their potential co-occurrence and semantic as- sociations within the entity category, remains largely uncharted. In this study, we pro- pose an approach to address this challenge. We construct a graph by extracting entity nodes, enhancing the contextual representation of unseen entities by leveraging the 1-hop surrounding nodes of associated entities. Furthermore, we introduce a named entity tag prediction task to address the issue of unseen entities not present in the constructed graph. We conduct our experiments on an open dataset Wizard of Wikipedia and the empirical results indicate that our approach outperforms the state-of-the-art approaches on Wizard of Wikipedia.

Statistics

   Article View: 229
   PDF Downloaded: 23