Knowledge Graphs (KGs) have applications in many domains such as
Finance, Manufacturing, and Healthcare. While recent efforts have led
to the construction of large KGs, their content is far from complete and
sometimes includes invalid statements. Therefore, it is crucial to
refine these KGs to enhance their coverage and accuracy via KG
completion and KG validation. Both of these tasks are particularly
challenging when performed under the open world assumption. It is also
vital to provide human-comprehensible explanations for such refinements,
so that humans have trust in the KG quality. Moreover, exploring KGs, by
search and browsing, is essential for users to understand the KG value
and limitations towards down-stream applications. However, enabling KG
exploration is a challenge task given the large size of existing KGs
This dissertation tackles the challenges of KG refinement and KG
exploration by combining symbolic reasoning over the KG with other
techniques such as KG embedding models and text mining. Through such a
combination, the proposed methods handle the complex nature of KGs and
provide human-understandable output. Concretely, we introduce methods to
tackle KG incompleteness by learning exception-aware rules over the
existing KG. Learned rules are then used for accurately inferring
missing links in the KG. Furthermore, we propose a framework for tracing
evidence for supporting (refuting) KG facts from both KG and text.
Extracted evidence is used to assess the validity of KG facts. Finally,
to facilitate KG exploration, we introduce a method that combines KG
embeddings with rule mining to compute informative entity clusters with
explanations.