Ontologies have long been employed in the life sciences to formally represent and reason over domain knowledge and they are employed in almost every major biological database. Recently, ontologies are increasingly being used to provide background knowledge in similarity-based analysis and machine learning models. The methods employed to combine ontologies and machine learning are still novel and actively being developed. We provide an overview over the methods that use ontologies to compute similarity and incorporate them in machine learning methods; in particular, we outline how semantic similarity measures and ontology embeddings can exploit the background knowledge in ontologies and how ontologies can provide constraints that improve machine learning models. The methods and experiments we describe are available as a set of executable notebooks, and we also provide a set of slides and additional resources at https://github.com/bio-ontology-research-group/machine-learning-with-ontologies.

/pdf/semantic-similarity-and-machine-learning-with-ontologies-4adxg9apmn.pdf

Semantic similarity and machine learning with ontologies

EL is a simple tractable Description Logic that features conjunctions and exis- tential restrictions. Due to its favorable computational properties and relevance to existing ontologies, EL has become the language of choice for terminological reasoning in biomed- ical applications, and has formed the basis of the OWL EL profile of the Web ontology language OWL. This paper describes ELK—a high performance reasoner for OWL EL ontologies—and details various aspects from theory to implementation that make ELK one of the most competitive reasoning systems for EL ontologies available today.

https://tpbin.com/uploads/subjects/78ddf721-e6bb-4862-b90a-94abf49b48ac.pdf

From Polynomial Procedures to Efficient Reasoning with EL Ontologies

Semantic embedding of knowledge graphs has been widely studied and used for prediction and statistical analysis tasks across various domains such as Natural Language Processing and the Semantic Web. However, less attention has been paid to developing robust methods for embedding OWL (Web Ontology Language) ontologies which can express a much wider range of semantics than knowledge graphs and have been widely adopted in domains such as bioinformatics. In this paper, we propose a random walk and word embedding based ontology embedding method named OWL2Vec*, which encodes the semantics of an OWL ontology by taking into account its graph structure, lexical information and logical constructors. Our empirical evaluation with three real world datasets suggests that OWL2Vec* benefits from these three different aspects of an ontology in class membership prediction and class subsumption prediction tasks. Furthermore, OWL2Vec* often significantly outperforms the state-of-the-art methods in our experiments.

OWL2Vec*: Embedding of OWL Ontologies

Semantic embedding of knowledge graphs has been widely studied and used for prediction and statistical analysis tasks across various domains such as Natural Language Processing and the Semantic Web. However, less attention has been paid to developing robust methods for embedding OWL (Web Ontology Language) ontologies, which contain richer semantic information than plain knowledge graphs, and have been widely adopted in domains such as bioinformatics. In this paper, we propose a random walk and word embedding based ontology embedding method named OWL2Vec*, which encodes the semantics of an OWL ontology by taking into account its graph structure, lexical information and logical constructors. Our empirical evaluation with three real world datasets suggests that OWL2Vec* benefits from these three different aspects of an ontology in class membership prediction and class subsumption prediction tasks. Furthermore, OWL2Vec* often significantly outperforms the state-of-the-art methods in our experiments.

/pdf/owl2vec-embedding-of-owl-ontologies-4hm602mqsl.pdf

In the past decade, we have witnessed a significant adoption of ontologies in a variety of fields including biomedicine, finance, engineering, law, and cultural heritage. The ontology engineering field has been strengthened by the adoption of several standards pertaining to ontologies, by the development or extension of ontology building tools, and by a wider recognition of the importance of standardized vocabularies and formalized semantics. Research into ontology engineering has also produced methods and tools that are used more and more in production settings. Despite all these advancements, ontology engineering is still a difficult process, and many challenges still remain to be solved. This paper gives an overview of how the ontology engineering field has evolved in the last decade and discusses some of the unsolved issues and opportunities for future research.

Ontology engineering: Current state, challenges, and future directions

In this paper, we present a preliminary study to compute embeddings for OWL 2 ontologies by projecting the ontology axioms into a graph and performing (random) walks over the ontology graph to create a corpus of sentences. This corpus is then given to a neural language model to create concept embeddings. The conducted preliminary evaluation shows promising results.

/pdf/embedding-owl-ontologies-with-owl2vec-32ba48xhli.pdf

Embedding OWL ontologies with OWL2Vec

While semantic technologies and the use of ontologies is one of the proposed solutions to the heterogeneity problem in database systems, the independent generation of ontologies leaves us with another heterogeneity problem. We need to match ontologies to allow active co-operation of different systems. This matching can often not be done by human effort alone because of the size and complexity of the ontologies. Thus automatic ontology matching systems are indispensable. We evaluate the usefulness of embeddings, more specifically ontology embeddings both for analytical tasks and for ontology alignment tasks. We further propose OWL2Vec as a framework to create such embeddings. The embeddings are created by projecting the ontology to an RDFgraph, conducting a series of walks on the graph and collect them in a walks document. Natural language embeddings systems such as word2vec are used to train the embeddings. To leverage ontology matching, we propose to create joint embeddings in the same vector space for two (or more ontologies). The user must then provide the system with anchors. The results are promising for analytical tasks such as clustering and partitioning of the ontology into related concepts. We demonstrate that by using the embeddings and the provided anchors, it is possible to find new mappings between two ontologies. We also demonstrate that the structural similarities provided by the embeddings can give ontology matching systems an additional similarity score that could help such systems to decide if a potential mapping is valid.

Semantic Embeddings for OWL 2 Ontologies

Requirements are critical components in the industry, describing qualities that a product or a service needs to have. Most requirements are only available as natural language text embedded in a document. Working with textual requirements is getting increasingly difficult due to the growing number of requirements, and having the requirements available as structured data would be beneficial. However, the work required for the translation of natural language requirements into structured data is daunting. Thus, we need tools to aid in this process. In this Ph.D. project, we propose to use state-of-the-art knowledge extraction techniques and develop novel methods to identify the terms and relationships in a requirement and align them with an existing domain-ontology. To achieve this goal, we must overcome the difficulties in working with both domain-specific technical corpora and ontologies. Furthermore, existing tools and NLP models must be adapted to the domain.

Ole Magnus Holter

Papers

OWL2Vec*: Embedding of OWL Ontologies

OWL2Vec*: Embedding of OWL Ontologies

Embedding OWL ontologies with OWL2Vec

Semantic Embeddings for OWL 2 Ontologies

Semantic Parsing of Textual Requirements