Masters Thesis Defense “Transfer Learning for Knowledge Graph Reasoning using Normalization” By BhargavaCharan Reddy Kaithi

Committee:  Drs. Pascal Hitzler, Advisor, Mateen Rizki, and Yong Pei

ABSTRACT:

In recent years, the research in deep learning and knowledge engineering has made a wide impact on the data and knowledge representations. The research in knowledge engineering has frequently focused on modeling the high level human cognitive abilities, such as reasoning, making inferences, and validation.

Semantic Web Technologies and Deep Learning have an interest in creating intelligent artifacts. Deep learning is a set of machine learning algorithms that attempt to model data representations through many layers of non-linear transformations. Deep learning is increasingly employed to analyze various knowledge representations mentioned in Semantic Web and provides better results for Semantic Web Reasoning and querying.

Researchers at Data Semantic Laboratory (DaSe lab) have developed a method to train a deep learning model which is based on End-to-End memory network over RDF knowledge graphs which can be able to perform reasoning over new RDF graph with the help of triple normalization with high precision and recall when compared to traditional deductive algorithms. Researchers have also found out that it’s 40 times faster to train than the non-normalized model on a dataset which they have performed experiments on. They have created efficient model capable of transferring its reasoning ability (by applying normalization) from one domain to another without any re/pre-training or fine-tunning over new domain which constitutes Transfer learning.

In this thesis, we are testing this Normalized embedding approach on the research which is done by Bassem Makni and James Hendler “Deep Learning for Noise-tolerant RDFS reasoning” [a]. The main limitation of their approach is that the training is done on a dataset that uses only one ontology for the inference. In order to overcome this limitation, we are proposing transfer learning process by adding the normalization approach created by DaSe Lab Researchers for reasoning over different ontologies/domains.