A great piece from Harry Chen via his Geospatial Semantic Web Blog. I have nothing to add to this bar: Amen! Enjoy the rest of his post below:
Hiding Ontology from the Semantic Web Users: "
Ontology is a key foundation of the Semantic Web. Without ontology, it will be difficult for applications to share knowledge and reason over information that is published on the Web. However, it is a serious mistake to think that the Semantic Web is simply a collection of ontologies.
Last week I was invited to be on a panel discussion at the Humans and the Semantic Web Workshop. I talked a bit about the Geospatial Semantic Web and its associated research issues. Overall the workshop went very well. You can read about the notes from the workshop here.
New Thinkings
Some of my new thinkings after the workshop are as the follows.
- People, especially those who are new to the Semantic Web, have put too much emphasis on developing ontologies and not enough emphasis on developing application functions.
- While ontology languages such RDF and OWL are important part of the current Semantic Web development, it’s a mistake to build Semantic Web applications that assume that average users are fluent in those languages.
- Many people seem to have forgotten that building Semantic Web applications don’t have start with ontology development. It’s a good idea to start with ontology reuse — i.e. reuse ontologies that have already been developed even if they don’t meet every single requirements of the application.
- There is no excuse to build ‘crappy’ UI just because developing Semantic Web applications are challenging.
Hide Low-Level Details from the Semantic Web Users
I was asked the question, ‘What’re user-related issues that Semantic Web developers must pay attention to?’ I think building Semantic Web applications are similar to building database applications. Few things we can learn from our past experience in building database applications.
When building database-driven applications, we store information in SQL databases, and we use SQL to access, manipulate, and manage this information. When building Semantic Web applications, we express ontologies and information in RDF, and use RDF query languages (e.g. SPARQL) to access and manipulate this information.
When building database-driven applications, we hide complexity from the end-users. For example, we almost never expose raw SQL statements to the end users, or ask users to process the raw result sets returned from an SQL engine. We always provide intuitive interfaces for accessing and representing information.
When building Semantic Web applications, we should also hide complexity from the end-users. Users shouldn’t need to see or edit RDF statements. Users shouldn’t need to be fluent in SPARQL queries or able parse graphs that are returned by a SPARQL engine.
Concluding Remarks
Semantic Web developers should spend more time on building functional capabilities that solve real world problems and improve people’s productivity. It’s important to remember that ‘the Semantic Web != ontologies‘.
"