Semantic Web in the Enterprise
We saw in the previous lesson that the primary use of Semantic Web on the (World Wide) Web is for publishing and consuming structured, linked data. Just as HTML was adopted inside enterprise IT for purposes that went far beyond intranet Web pages (e.g. software documentation or email formatting), Semantic Web technologies are being used today in companies for a wide range of applications. In this lesson, we’ll look at some examples of how Semantic Web technologies can be used that go beyond publishing data on the Web.
Within an enterprise, Semantic Web technologies are a technology stack that can be chosen to implement solutions and applications, just as one might consider using XML or relational technologies. People choose to use semantic technologies for a variety of reasons, some of which are highlighted here.
Agile Data Integration
One of the most common ways that enterprises are leveraging Semantic Web technologies is as a cheaper, incremental, and more agile approach to data integration.
Large organizations struggle all the time to launch large data warehouse projects or EAI/EII efforts in order to provide employees with unified views of the information they need. Fundamentally, traditional data integration approaches suffer from the very same lack of flexibility that semantic technologies are designed to overcome.
Note: We’ll compare Semantic Web technologies with the relational technologies of the data warehouse and EAI/EII world in a future lesson dedicated to it.
Consider three examples:
1. The Department of Defense is linking together scores of disparate information systems by overlaying them with a semantic model. The DoD’s Business Mission Area issued a mandate for the use of Semantic Web technologies as the foundational architecture for new integration projects in 2011, and they’ve since seen the time it takes to bring a new enterprise system into the fold from 6 to 9 months or longer to under 90 days.
2. The pharmaceutical company Biogen Idec applies Semantic Web technologies to address data integration challenges involved in collecting and understanding supply chain metrics generated across the 30 manufacturing partners with whom they do business. For Biogen, doing supply-chain data integration via semantic technologies means that they can integrate data far more easily than before and can begin tracking new metrics and KPIs to get a better understanding of their overall supply chain. You can read more about Biogen’s data integration use case in our lesson on Example Semantic Web Applications.
3. Amdocs is using Semantic Web technologies to give large telecom companies a far more complete and useful view of their customers than they’ve ever had before. For Amdocs, semantic technologies provide the perfect infrastructure for pulling together information from voice-response systems, billing and payment systems, support systems, sales and marketing systems, and more to provide the full context of a customer’s interactions with a telecom company as its needed. The flexibility of the underlying model means that they can base continuous process change, decision models, and proactive alerts on this unified view of customers without knowing ahead of time what their full requirements are.
In these examples, the primary reason semantic technologies were considered was their flexibility compared to traditional tools. As we covered in Applying the Semantic Web: Two Camps, this pragmatic advantage compared to other tools can enable solutions to be evolved more easily.
Some organizations are turning to Semantic Web technologies to help tackle the classic needle in a haystack problem. Generally, the needle-in-a-haystack problem has two dimensions. The first is getting all relevant data in one place, which is an example of the agile data integration challenge that we’ve already considered.
The other dimension of this challenge is classifying the information in the haystack well enough to facilitate finding the needles, a significant challenge in today’s world of Big Data. Companies and other organizations are combining the expressivity of ontologies with support for automated reasoning (inference) as they apply Semantic Web technologies to problems of information classification.
The use of semantic technologies for information classification can range from improved capabilities to (manually) tag and retrieve information through to fully automated information discovery. At the former end of this spectrum, libraries—including the Library of Congress, British Library, and Stanford University—and related organizations—such as Europeana—have begun to represent their vast catalogs of bibliographic information with Semantic Web technologies.
At the other end of the spectrum, reasoning is being applied to large data sets described with Semantic Web ontologies in order to automate the process of locating key bits of information. In the case of the SAPPHIRE project from the University of Texas Health Science Center, this means combining and classifying clinical data from various institutions to identify potential outbreaks of the flu. In the case of a project submitted to the FDA by the NCE CECR Centre of Excellence for Prevention of Organ Failure (PROOF), this means identifying and classifying patients who risk immune rejection of organ transplants before they exhibit symptoms.
As we covered in Applying the Semantic Web: Two Camps, the group that focuses on the classification aspects of Semantic Web technologies has a lot in common with the AI community. There are many active research projects both in academia and in corporate R&D groups focused on improving this capability.
Dynamic Content Management
The BBC chose to use Semantic Web technologies to drive the infrastructure that they used to produce and publish their website for the 2010 World Cup. This website combined manual tagging, ontology-driven reasoning and
automated topic aggregation to produce a highly dynamic, interactive, and—ultimately—more useful site. Traditional content management approaches would never have reached the richness of information available on the
BBC’s site: the World Cup site included 700 topical index pages, which is more than the entire rest of the BBC sports site, combined.
For the BBC, the value of Semantic Web technologies in the enterprise is in the flexibility and automation that they bring to content management in two areas:
1. The flexibility allowed them to produce an information-rich site at many levels of aggregation (e.g. player, team, geography, group, etc.) without employing a large fleet of editors to curate and interlink the site’s content.
2. They leveraged the expressivity of Semantic Web ontologies to help automate the tagging process by taking advantage of taxonomy structures and natural semantic relationships between concepts.
The BBC has blogged about their experiences with Semantic Web technology a couple of times, and this use case is also described in more detail elsewhere in Semantic University.
Another form of dynamic content management involving Semantic Web technologies that companies are employing is Semantic Wikis, which we’ll cover in a future lesson.
All of the Above
Of course, many uses of Semantic Web technologies within an enterprise don’t clearly fall into one category or another. For example, it’s pretty common for an application that’s is mainly geared at agile data management to include at least some classification, and also some kind of publishing capability.
Furthermore, these aren’t the only categories: some applications focus on search, on question answering, on modeling, and more. Compared to the heavily data-oriented uses of the Semantic Web on the Web today, usage of Semantic Web technologies within the enterprise offer a range of possibilities to derive practical value today in a variety of application areas. For more on choosing good places to apply these technologies, see the lesson on What Makes a Good Semantic Web Application.