Looking at the existing investments into information technology, why should you invest into Semantic Web? What are the key benefits I will get? And who can give me authentic data about experiences? In this research note, I reviewed seventeen case studies of applied Semantic Web in industry settings. These case studies were submitted by the respective authors to the SWEO interest group of W3C. By taking the 89 key benefits mentioned in the individual examples and integrating them into a classification scheme, we can see the important benefits of Semantic Web and the not-so-important. I also identify a possible area for growth. The presented data is not peer reviewed and sloppy, but can still give you a guide to using the Semantic web.
At ISWC2005, one talk presented a slide about "what are the key reasons to use Semantic Web". Sadly, I lost the reference and the paper did not list them. Interestingly, there is not much data available about the actual benefits of deploying Semantic Web technology. To create such overviews, we can either do interviews and questionnaires or look at documented case studies.
The Semantic Web Education and Outreach SWEO Interest Group of the W3C did collect and publish 27 case studies and 10 use cases (numbers as of 22.4.2008). The case studies report of deployed systems, therefore I concentrate of them. As methodology, I copy/pasted all key benefits into a spreadsheet and mapped them to classifications I made up on the run. I added explanatory text about the categories by copy/pasting statements by the authors. Then I sorted the results, showing the top-named benefits first. One argument from a use case could map to multiple of my categories (i.e. when a benefit touches both a business profit and is about data integration, it will be added to both categories).
The correct scientific method to do such study would be to use questionnaires and to check if the statements given by the authors are true, so the scientific value of this note is questionable, but I love controversy.
|some descriptive words
|customer satisfaction, shareholder value, user work support
|data integration accross sources
|A consistent data model to build upon. integrate content from different organizations /providers/ departments, disparate data sources, legacy data
|software uses the terminology of the user, vocabulary, synonyms, multilinguality
|accessible for users
|generate website make data accessible to users
|navigation, related items, serendipity
|discover hidden information, navigate accross data boundaries, "what works with what"
|incorporate unanticipated data sets, new requirements do not change code, dynamic adaptability
|existing data can be repurposed, unprecedented use cases, conserve culture standards
|inference owl reasoning formal
|cheap implementation lightweight
|improve content quality
|web of linked open data
The source data is available as spreadsheet here. Looking into the spreadsheet, you can replicate the decisions I made when assigning the statements to classes.
I sum up my conclusions in bullets, so you can easily copy-paste them into your next presentation slides.
Focussing on the last point, I want to stress that given the Linking Open Data project (also kind-of-initiated by SWEO), you now have a very good overview of the publicly available Semantic Web data. People want automated features and to reduce workload on users, you must reuse the data that is out there and I would give the prophecy that your business value will increase if you reuse this data.
The call to submit more SWEO case studies is kept open, if you submit new data I will try to keep this document updated. Otherwise, I do not have the time to invest much into to this document, so I will not incorporate all your feedback, but appreciate comments on my weblog post:
The CSS was created by Richard Cyganiak for our mutual publication "Cool URIs for the Semantic Web". I re-use it here. All data was taken from the publicly available SWEO case studies. Some statements from the case studies were copied literally and were created by their respective authors.
Copyright Leo Sauermann, 2008, all rights reserved, blogging encouraged.