We present a family of novel methods for embedding knowledge graphs into real-valued tensors. These tensor-based embeddings capture the ordered relations that are typical in the knowledge graphs represented by semantic web languages like RDF. Unlike many previous models, our methods can easily use prior background knowledge provided by users or extracted automatically from existing knowledge graphs. In addition to providing more robust methods for knowledge graph embedding, we provide a provably-convergent, linear tensor factorization algorithm. We demonstrate the efficacy of our models for the task of predicting new facts across eight different knowledge graphs, achieving between 5% and 50% relative improvement over existing state-of-the-art knowledge graph embedding techniques. Our empirical evaluation shows that all of the tensor decomposition models perform well when the average degree of an entity in a graph is high, with constraint-based models doing better on graphs with a small number of highly similar relations and regularization-based models dominating for graphs with relations of varying degrees of similarity.

Multi-relational data, like knowledge graphs, are generated from multiple data sources by extracting entities and their relationships. We often want to include inferred, implicit or likely relationships that are not explicitly stated, which can be viewed as link-prediction in a graph. Tensor decomposition models have been shown to produce state-of-the-art results in link-prediction tasks. We describe a simple but novel extension to an existing tensor decomposition model to predict missing links using similarity among tensor slices, as opposed to an existing tensor decomposition models which assumes each slice to contribute equally in predicting links. Our extended model performs better than the original tensor decomposition and the non-negative tensor decomposition variant of it in an evaluation on several datasets.

We describe our system used in the 2018 FEVER shared task. The system employed a frame-based information retrieval approach to select Wikipedia sentences providing evidence and used a two-layer multilayer perceptron to classify a claim as correct or not. Our submission achieved a score of 0.3966 on the Evidence F1 metric with accuracy of 44.79%, and FEVER score of 0.2628 F1 points.

We describe the systems developed by the UMBC team for 2018 SemEval Task 8, Se-cureNLP (Semantic Extraction from Cyberse-cUrity REports using Natural Language Processing). We participated in three of the subtasks: (1) classifying sentences as being relevant or irrelevant to malware, (2) predicting token labels for sentences, and (4) predicting attribute labels from the Malware Attribute Enumeration and Characterization vocabulary for defining malware characteristics. We achieved F1 scores of 50.34/18.0 (dev/test), 22.23 (testdata), and 31.98 (test-data) for Task1, Task2 and Task2 respectively. We also make our cybersecurity embeddings publicly available at https://bit.ly/cybr2vec.

No abstract