Question answering over knowledge graphs

The web, as a primary information resource, is holding a massive amount of structured data in the form of knowledge graphs (KGs). KGs include facts ranging from hundreds to billions in various domains such as tourism, healthcare, manufacturing, and marketing. The standard way to access facts stored in KGs is using query languages. However, end-users find access difficult due to the complexity of the query languages and the structure of KGs. To make KGs accessible for end-users, question-answering systems allow the end-users to interact in a natural way with the web and meet their information needs while they remain unaware of the underlying structure of the KGs.

In this work, we present two approaches for developing question-answering systems across small and medium-scale KGs and huge KGs. Such systems enable end-users to ask their questions and receive precise answers. We apply our solutions to a small and medium-scale KG in the domain of tourism and to an open-domain and huge KG.