SEMANTiCS 2017: An Interview with Ontotext’s CEO Atanas Kiryakov

The SEMANTiCS conference series is organized by Semantic Web Company and local partners since 2005. It has a longstanding tradition of bringing together enthusiasts from around the world to present best practices, panels and papers on Semantic Technology. Learn about Ontotext beginnings, and its vision for Linked Data and Semantic Technology in this interview with Ontotext’s CEO Atanas Kiryakov, which was taken before this year’s event on September 11-14 in Amsterdam. It was published originally on August 28th at www.semantics.cc .

SEMANTiCS 2017: Can you tell something about your work and research focus?

Atanas Kiryakov: In the 90’s I was involved in researching knowledge representation, upper-level ontologies, lexical semantics, NLP, information retrieval. I started Ontotext in the year 2000 and we became Semantic Web pioneers. We developed the concept for semantic annotation, indexing and retrieval, where big Knowledge Graphs are used to develop cognitive reading machines.

Such machines have sufficient knowledge about the domain and the language to properly recognize concepts and entities, and extract new facts. Today, this is the Ontotext Platform, which is typically used for automated tagging, semantic search and recommendation. To build such machines, we needed a good semantic graph database. We started using the existing ones but quickly figured out that they were immature. So, we had to develop one on our own – this is how OWLIM started, now its name is GraphDB.

SEMANTiCS 2017: Which trends and challenges do you see for Linked Data and Semantic Technology and why are they important for Ontotext?

Atanas Kiryakov: We see a wave of big enterprises that start managing their master data and metadata using Semantic Web standards. We have plenty of clients that have Linked Data platforms or something similar.

Sometimes they use open data, but most of the time this is not the main reason. Big enterprises need to share and publish data internally. They adopt Linked Data standards because they are designed for publishing and linking information across different sources – this is an important difference when compared to property graphs. RDF triplestores implement W3C standards such as RDF(S), OWL and SPARQL.

What is even more important for master data management and metadata publishing is that triplestores use global identifiers and robust schema languages. Taken together, these factors establish RDF databases as the paradigm of choice for metadata and master data management in the biggest international companies. This is good for Ontotext as a developer of GraphDB – the leading semantic graph database.

We also see a growing adoption of Semantic Technology for content management. Publishers and other organizations that own good quality content use metadata to facilitate its management and monetization. They publish metadata in order to promote their content and to make it easier for the readers to find it and to explore it. This way they counter-fight the overwhelming volumes of cheap and shallow content. This is exactly the case with Springer Nature, who chose GraphDB for their SciGraph project.

Finally, there are emerging tendencies for the adoption of open data about people, organizations and locations – we call it POL data (from the abbreviation). There are growing volumes of such data, but we also see a growing number of sources or publishers.

DBpedia has been around for quite a while, but its coverage of legal entities is very limited, about 0.1% of all organizations in the world. Sources like OpenCorporates provide data for more than 100 million organizations, which is a game changer. And there are new sources of such information like GLEI – the Global Legal Entity Identifiers initiative, where volumes are also growing quickly, partly because of regulations that require organizations to subscribe and get a global LEI identifier.

You can think of this open data like a sort of global master data. Enterprises can use this data to complement their master data about, say, clients. This saves them the effort to collect and update broad information about these entities themselves. What’s the point of maintaining internal nomenclatures of locations, when there is GeoNames that covers all locations on Earth in a fairly systematic manner? Using open data as master data in the enterprise has its own challenges of course, but Ontotext has been working on this for years and we already have solutions.

We are involved in the EU project euBusinessGraph, which is developing a highly interconnected graph of Europe-wide company-related information, as well as a data marketplace for such information.

SEMANTiCS 2017: What are your expectations about Semantics 2017 in Amsterdam, what makes it special for Ontotext as a company?

Atanas Kiryakov: We have attended Semantics conference for several years in a row. This is the best European event to meet other vendors of Semantic Technology and outstanding early adopters across all sectors and we are a proud sponsor of the conference.