In 2018 Ontotext grew up as business and technology. We are closing up the year with wonderful financial results, but there are even more important and more exciting developments.
We see more mature demand – customers who use semantic technology adequately to solve real business problems. More and more often we see inquiries for GraphDB from multinational corporations who have built concise AI architectures around knowledge graphs, outlined technical requirements, evaluated several engines and are ready to onboard a semantic graph database. Recent examples are some of the biggest automakers and aircraft manufacturers. We keep developing our vertical expertise and offerings to better address market-proven needs in company intelligence and healthcare.
To be able to target new use cases and expand the added value we offer, we have integrated new types of technology in our products, e.g., Semantic Vectors for statistical inference based on machine learning and document store for large-scale metadata management. To accelerate our R&D further, we have secured more than €3M external funding.
This makes a great starting point for our new big mission – to become the heart of the Artificial Intelligence (AI) plans of Sirma Group, our majority shareholder. Back in April 2017, Sirma Group announced its Corporate Strategy 2022 towards enterprise AI. The ambition of Sirma is to become a prominent vendor of enterprise AI technology, which offers an end-to-end AI platform (SENPAI) and has the capacity and the partner ecosystem to deliver it worldwide.
In the rest of this blog post, I elaborate on the above topics. The post is decorated with pictures of our company values – the artwork of our designer Dilyana Angelova
This year we have continued to expand the functionalities of our signature semantic graph database, GraphDB, with a new range of capabilities.
One of the most promising new features is the Semantic Vectors package integrated as a plugin for search based on concept similarity in knowledge graphs. We, as humans, determine the similarity between texts based on the similarity of the composing words and the cognitive associations and relationships encoded in our brains. Now, this new plugin enriches the RDF graph with semantic similarity indices, based on a highly scalable vector space model. This vector representation mechanism, which became popular as “embedding”, is special – it is derived through machine learning techniques that encode similar concepts into mathematically “similar” vectors. Thanks to it, users can perform statistical inference and get more results based on the matching of semantically close concepts.
Another new plugin is the MongoDB connector for GraphDB. MongoDB is the document database with the biggest developer community. Its indices are designed for fast storage and retrieval of documents and objects, unlike graph databases, which are most efficient in analyzing relationships, detecting patterns and inferring new facts. Document databases allow a more flexible approach to data and data changes and guarantee better scalability and performance for updates and object retrieval. Most importantly, the combination of a general purpose approach and the explicit requirement to introduce indices for specific types of queries makes MongoDB more suitable for very large sets of data, documents, annotations, etc.
GraphDB got proven through the years as a mature and resilient engine, powering business critical operations for some the most knowledge-intensive enterprises on Earth. With these new additions to its ecosystem of special-purpose indices and connectors to other systems, GraphDB is ready to serve as a central component in a very wide range of enterprise content management and data management architectures and to handle efficiently various analytics and transactional loads. To serve better the entire lifecycle of the data, we have also developed a new mechanism for data extraction, transformation and loading (ETL), based on Apache SPARK. It allows for parallelizable, and thus horizontally scalable, ingestion, cleaning and normalization of data from diverse sources. Now GraphDB is ready to take care of the data layer in SENPAI.
During the year, we have successfully implemented a new delivery methodology based on the Agile approach. This implementation was accompanied by some intensive team training. As a consequence, the efficiency has increased to more than 80% billability.
The matured process and delivery discipline led us to reaching annual revenue of 100 000 euro per productive employee. This is a significant achievement, given that this average value includes not only professional services and operations personnel, but also teams dedicated to product development, research and training. The improved efficiency had a direct positive impact on profitability and cash flow.
The offerings of our operations team are moving up the value chain. In 2018 Ontotext signed several contracts to assume more responsibility for entire systems and IT functions, rather than just to provide third level technical support. Such offerings include managed data and analytics services, where we take the responsibility to meet key performance indicators (KPI), determined by the business needs. On top of the standard operational KPIs (e.g., response time and availability), such managed services involve guarantees for data quality and text analysis accuracy in tailor-made information architectures, which are updated with new data coming from multiple external sources. E.g., a knowledge graph of company data, which integrates and interlinks data from 5 different data vendors.
With this increased project management and operations capacity, we were able to improve our Customers and Employees satisfaction and project predictability. Our ambition is to continue improving our efficiency next year.
In November 2018, we were happy to mark an important milestone in our cooperation with Fujitsu - the World’s 7th largest IT services provider and No.1 in Japan. Fujitsu Technology Solutions is now using Ontotext GraphDB to deliver projects with the Ministry of the Interior of Spain in the field of National Security.
After InfoSys, Atos Origin and NTT Data, this is the fourth Top 10 IT Services provider to adopt Ontotext technology for its Artificial Intelligence platform. Partnering with global IT service providers, consultants and system integrators is a key part of Ontotext’s and Sirma’s strategy for business development and delivery of business solutions to big enterprises and governments around the world.
In 2018, four proposals with the involvement of Ontotext got funding by the European Commission. The total value of the research grants for Ontotext in these projects is €1.8M.
Two of them have already started. The first is CIMA (Intelligent Matching and Linking of Company Data), which focuses on harmonizing data through semantic representation and integration. It is also developing methods for semantic matching, linking and entity extraction. The second is WeVerify (Wider and Enhanced Verification for You), which aims to expose fake content through cross-modal verification, social network analysis, micro- targeted debunking and a blockchain-based public database of known fakes.
Two more will start in January. EXA MODE (EXtreme-scale Analytics via Multimodal Ontology Discovery & Enhancement) is a Big Data project for the Healthcare. It involves the development of extreme analytic methods for decision-making by hospitals. InnoRate, on the other hand, (Data-driven tools for supporting and improving the decision-making processes of investors for financing innovative SMEs) has the ambition to disrupt the largely risk-averse financial sector in Europe and to enhance the innovation capacity of high growth technology sectors and SMEs.
TRR (Tracking of Research Results) as another project with Ontotext’s involvement, which started in October. It is not a research project, but it has a direct bearing on several research projects. TRR aims to enable the policy makers in research and innovation to analyze the outcomes of research projects. The scope of the project includes the identification and tracing of information about research results, new products based on them,
All these projects bring the total amount of financing secured to support Ontotext’s strategic R&D plans to €3.1M for 2018 and the following years. The financing is allocated by strategic verticals as follows: €1.9M for Market Intelligence and Publishing and €1.2M for Healthcare and Life Sciences. This will bolster the development of vertical knowledge models and specific applications for Sirma’s enterprise AI platform SENPAI.
History likes cycles. Sirma Group established in 1992 as a Sirma AI Ltd. А quarter century ago we developed several expert systems for government clients in Canada and elsewhere. E.g. expert systems based on Boolean Constraints Propagation Networks for tax advice and for alerts in case of pollution disasters. In the mid ’90s, however, AI went out of fashion – the marketing teams of numerous aggressive startups boosted the expectations to a level that the technology was not ready to deliver. Sirma diversified its business and today it is one of the biggest IT groups in Bulgaria. Sirma Group is publicly traded on the Sofia Stock Exchange (SKK:BLG).
I started Ontotext in the year 2000 as an R&D lab within Sirma and in 2008 we spun it off to accommodate funding for productization of semantic technology. Later, the attitude towards AI changed again. In the recent years, AI technology has developed and has proven its potential to revolutionize various industries. Still, most of the AI technology vendors are geared towards consumer applications, e.g., smart home assistants, rather than improving the efficiency of the enterprises. Sirma spotted this opportunity and developed its Corporate Strategy 2022 towards enterprise Artificial Intelligence (AI).
Central to Sirma’s plans is Ontotext: its products, its know-how in the development and marketing of world-leading technology, its R&D capacity and its track record in delivering enterprise AI projects. Executing this strategy, Sirma Group purchased the shares that VC fund NEVEQ had in Ontotext. Sirma also started a secondary public offering (SPO) to collect extra funding to boost the development of the SENPAI platform and to support quicker business expansion.
As the next step in this process, Sirma will consolidate the R&D capacity and the intellectual property of the group related to AI. The plan involves merging Ontotext into “Sirma AI” JSC – a wholly owned subsidiary of Sirma Group. By the end of 2018, Sirma AI will take over Ontotext as an entire business enterprise, including its management, staff, contracts and obligations. Sirma AI will develop further GraphDB and Ontotext Platform and will grow the SENPAI platform around them. I am having the honor to be the CEO of Sirma AI.
For 18 years Ontotext has implemented the full cycle of innovation around the vision of using big knowledge graphs for text analysis, data linking and analytics. We have participated in more than 30 research projects and have written hundreds of scientific papers, which have thousands of citations. We have come up with inventions and have attracted VC funding. We have matured our products into robust enterprise IT infrastructure running business critical systems. Finally, we have proven that this technology is commercially viable - we have created a business model that allows Ontotext to grow and to be profitable, competitive and sustainable.
It is time now to go to the next level - to evolve Ontotext and Sirma into big vendor that has the capacity to develop a broader enterprise AI platform and deliver it worldwide. Let wish ourselves good luck, brave hearts and cool heads!
135 "Tsarigradsko Shose" blvd., 1784 Sofia, Bulgaria tel: +359-2-976-8310
202 N 9th St, Suite 201B, Boise, ID 83702, USA Tel: (800) 701-3710 ext. 101
© Sirma Group 2018
I can answer questions like:
Who are the board of directors of Sirma?
Who is Tsvetan Alexiev?
Click Help for more examples