Today we continue our look ahead at what may be to come for The Semantic Web and related fields in 2015 (see here for our first story in this series):
Elisa Kendall, Thematix Partners partner and Member, Object Management Group Architecture Board & co-chair, and Ontology Definition Metamodel Revision Task Force:
One thing that I think will be increasingly important is the use of semantics for business vocabularies as part of an overall data governance program. The CMMI recently published a new Data Management Maturity Model, which includes recommendations for a comprehensive metadata strategy and business vocabulary as a component of a governance program. A number of the banks in the US and EU are taking a hard look at their data governance capabilities, in part because of this new model, but also due to some of the regulatory requirements on the horizon.
A couple of the major banks that we’re working with are developing ontologies to meet these requirements as a component of their data governance programs, and even more importantly to increase their internal business intelligence capabilities. They are extending the Financial Industry Business Ontology (FIBO) to get a significant leg up, but much of the work to aggregate data from the various systems they depend on is relatively new. With deadlines for compliance by January 2016, I would expect that we’ll see a lot of activity next year.
Also, several parallel working groups for FIBO Equities and Securities as well as Loans are kicking off in January, with major banks as well as Bloomberg participating, and I think we’ll hear more and more buzz about that as the ontologies from those efforts start to shape up.
I also think that there is more work to be done on the tooling front, especially for working with business people to visualize and collaborate on the vocabularies they need to support these requirements. We can cobble together a number of tools for development and deployment for various purposes and audiences now, far better than we could a few years ago, but it’s not seamless or nearly as intuitive as it should be. There are tools for terminology development and tools for ontology development, but a business vocabulary requires aspects of both, and what’s available currently doesn’t really address both needs.
Salar al Khafaji, Silk co-founder:
I still think the biggest problem actually goes back to the data and the wide variety of structures for the data in APIs. For us, the most important development in the space will be the emergence of more standardized API formats that will make it much easier to have APIs talk to each other or to query multiple APIs to pull together data that lives in other places and is dynamically updated. This would be a game changer — one API structure to rule them all.
JSON-LD is an interesting step in the right direction, for instance.
Andy Palmer, Tamr CEO and co-founder:
We are at the very beginning of the process of enterprise data simplification and unification [to the point where it can be usable in context in specific analytics use cases]. When we talk to some of the largest companies in the world, they tell us they have spent billions of dollars automating all their processes and optimizing them and creating data systems. Now the challenge is not generating data, but making data available in a unified way so that it’s consumable by all the people in a company as appropriate….A lot of the largest companies are changing their orientation from ‘I will buy from one vendor to unify all my data’ to embracing hundreds of thousands of legacy systems.
The only way to really get integrated data from those systems to support next-generation analytics in the infrastructure is providing heterogeneity of disparate systems, to modeling bottom up how all data from various systems is related…. I’ve advocated for the latest and greatest semantic technologies these last decades and it’s cool to see this stuff coming into mainstream [for these and other purposes].
Mark Montgomery, Kyield Founder & CEO:
In following from 2014 trends, among the most important risks to our species is not only misuse of advanced technology — certainly to include AI – but perhaps even more so at this stage of evolution is that cognitive bias, conflicted manipulation of information, and exploitation of ignorance could lead to a backlash that prevents what Yudkowsky coined as “Friendly AI.” This scenario would be plausible as one of the greatest risks under the umbrella of unintended consequences — ironically as AI wasn’t deployed soon enough to prevent the financial crisis, which led to broad joblessness, which could then well lead to a backlash against AI due to threat of job displacement, and so on. This is how chain reactions typically evolve that often does lead to crises, extend them, etc.
So my hope is that in 2015 we will collectively focus on the more plausible risks and benefits of AI and related technologies, including the risk of dominance from monopolies in the neural network economy, whether they be from traditional search, ecommerce, or emerging next generation technologies including AI. Cheers to a more peaceful and sustainable 2015.
Chris Nellligan, Dow Jones Professional Information Business VP and CTO:
In addition, as the interest and understanding of the Semantic Web continues to grow throughout corporations, new ontologies will be formed which will enable the joining of disparate datasets from multiple providers as well as ease integrating with such data sets. New vocabularies are materializing every month at an exponentially rapid rate. As for Factiva [its product that applies semantic tools to its content set to surface relevant information], the market can expect to see new semantic value added to its rich and ever expanding content set, with a strong focus on the quality and relevance of that semantic enrichment.
Matthew Sanchez, Cognitive Scale founder and CTO:
The Internet of Things (IoT) is one area that will complement cognitive computing. The ability to apply cognitive computing to IoT simply makes sense – we all want to better understand how to use and apply the massive amounts of information we have to create better products, deliver better service, and improve customer experiences.
Amit Sheth Ohio Center of Excellence in Knowledge-enabled Computing, LexisNexis Ohio Eminent Scholar, Wright State University:
- Graph >> RDF >> OWL. OWL and formal ontologies will have only limited and relatively niche applications because of the level of human effort involved in creating and maintaining ontologies, the availability of trained individuals, and the limited need for reasoning (e.g., for knowledge base consistency). RDF use will continue to be underwhelming, but we are in the middle of resolving issues in the performance and scalability of current RDF databases. Formal graph modeling for RDF is on its way, which will give a little bump to RDF usage. Further uptick in use of Linked Data, which has found significant use in some applications, will be hampered unless the issue of quality is handled. Relationships are at the heart of semantics and the Semantic Web; graphs, especially labeled graphs, are often sufficient to represent the relationships and relevant computations most sought (e.g., patterns and paths), give much better scalability and performance, and are familiar to a lot more programmers. So for the next few years, and perhaps longer, for Web scale applications, they will remain the only option.
- The most important technical challenge today in managing big data is variety (heterogeneity of data and diversity of data sources). The only effective way to handle heterogeneity is a semantic approach: Develop some form of vocabulary, knowledge base or ontology, and use semantic information extraction and annotate heterogeneous data to improve interoperability and integration.
- While aggressively using data mining, machine learning (ML), and NLP, the industry has significantly undervalued semantic approaches. As the value of a semantic approach to complement and improve these technologies becomes better understood, the growth will be hindered by a lack of trained software engineers with appreciation for the value of using background knowledge and Semantic Web expertise—lack of past demand has meant few people have really mastered semantic techniques.
- To sum this (and Sheth’s thoughts offered last week in our look back at 2014 here) in one thought: While there will be a marginal increase in use of Semantic Web standards (RDF/SPARQL, OWL/reasoning), more people will use semantics in the form of knowledge bases or ontologies and entity/concept graph network with relationships as edges.
Nova Spivack, technology futurist, serial entrepreneur, angel investor and CEO, Bottlenose:
I think we will see increasing momentum towards platformizing cognitive computing for real-time big data analytics. We’re working on that at Bottlenose and you should expect some interesting announcements from us in 2015. (Nova Spivack also authored a guest column on the recent past and the near future, which you can read here.)
Please feel free to weigh in with your own thoughts on what you’re looking forward to! (And as we mentioned last week, you also should head over to Dataversity.net for additional articles exploring the future for the Semantic Web, Cognitive Computing, NLP, Big Data and other affiliated areas.)