Knowledge Graph Embeddings.

TL;DR: Despite the importance of building large-scale KGs, their symbolic and logical frameworks are not flexible enough and can empower many semantic Web applications.

Abstract: With the growing popularity of multirelational data on the Web, knowledge graphs (KGs) have become a key data source in various application domains, such as Web search, question answering, and natural language understanding. In a typical KG such as Freebase Bollacker et al (2008) or Google’s Knowledge Graph Google (2014), entities are connected via relations. For example, Bern is capital of Switzerland. Formally, a popular approach to represent such relational data is to use the Resource Description Framework. It defines a fact as a triple (subject, predicate and object), which is also known as head, relation, and tail or (h,r, t) for short. Following the above example, the head, relation and tail are Bern, capitalOf and Switzerland, respectively. With a considerable number of entities and relations (e.g., Google’s Knowledge Graph has more than 18 billion of triples with 570 million of entities and 35,000 of relations by the end of 2014), KGs now become a valuable information source that can empower many semantic Web applications. Despite the importance of building large-scale KGs, their symbolic and logical frameworks are not flexible enough