One of Semantic Web’s Core Added Value

Frederick Giasson
If I ask the question: “What added value(s) does the Semantic Web brings on the table?”. So, what are the benefits that companies and organizations would get from using the Semantic Web? I am pretty sure that after asking this question, I would get answers such as:
  • You will instantly be able to traverse graphs of relationships
  • You will be able to infer facts (so create/persist new knowledge) from other existing facts
  • You will be able to check to make sure that your knowledge base is consistent and satisfiable
  • You will be able to modify your ontologies/vocabularies/schemas without impacting the description of your instance records or the usability of any software that use it (unlike relation databases)
  • And so on…

All these answers would be accurate. However, what if these answers would only be a part of the real added value that the Semantic Web brings on the table?

Note: when I refer to the “Semantic Web” on this blog post (and across all my writings), I refer to a set of technologies, techniques and concepts referred as the Semantic Web. So it is not a single thing, but a complete set of things that creates new ways of working with, and manipulating, information.

Strong of about 7 years of research and development of Semantic Web technologies that includes about 3 years of developing the Open Semantic Framework, that the biggest added value that I found from utilizing Semantic Web technologies is only partially related to these answers. In fact the biggest added value for me, as a developer can be resumed in one word:

PRODUCTIVITY

As simple as this. The biggest added value I gained from using and applying Semantic Web related technologies, techniques and concepts is an important increase in development, and data integration productivity.

Such productivity gain as to do with one of Semantic Web’s core attribute:

FLEXIBILITY

This is what I was suggesting in my latest blog post about Volkswagen’s use of the Open Semantic Framework: how Volkswagen uses the Open Semantic Framework to get flexibility that will lead to a gain in productivity to integrate, publish, and re-contextualize their data assets. The few gains that I listed above are part of the reason why the Semantic Web gives you flexibility that leads to an increase in productivity.

This same point as been re-affirmed today by Lee Feigenbaum in its latest blog post Saving Months, Not Milliseconds: Do More Faster with the Semantic Web:

Why is this? Ultimately, it’s because of the inherent flexibility of the Semantic Web data model (RDF). This flexibility has been described in many different ways. RDF relies on an adaptive, resilient schema (from Mike Bergman); it enables cooperation without coordination (from David Wood via Kendall Clark); it can be incrementally evolved; changes to one part of a system don’t require re-designs to the rest of the system. These are all dimensions of the same core flexibility of Semantic Web technologies, and it is this flexibility that lets you do things fast with the Semantic Web.

Warning: Productivity is not synonymous with simplicity

However, I would warn people that think that productivity gains are possible because semantic web technologies are simpler to use, manage and implement than other existing technologies.

It is certainly not the case, and I don’t think it will ever be. Semantic Web technologies, techniques and concepts are not easy to understand, and have a big learning curve. This is partly true because these techniques, technologies and concepts are relatively new in the field of the computer sciences, and because they are not fully understood, defined, implemented and used.

Volkswagen’s Use of structWSF in their Semantic Web Platform

Frederick Giasson

TribalDDB London, Volkswagen UK‘s partner, mentioned earlier this week that Volkswagen are using some parts of the Open Semantic Framework to develop the next generation of their online platform.

This story has been published by Jennifer Zaino’s in her article: Volkswagen: Das Auto Company is Das Semantic Web Company!

I can now talk about this project that uses some pieces of the framework that we have been developing for more than 3 years now.

The Objective

Volkswagen’s main objective behind the development of the next version of their Web platform started by improving their online search engine, but as William Greenly mentioned, it quickly became a strategic decision:

“So the objectives were about site search and improving it, but in the long-run it was always the idea to contextualize content, to facet content, to promote it in different contexts.”

The objective is to create a platform that gives them the flexibility to leverage all the data assets they own. This flexibility will help them to leverage the data assests they have to improve not only their search engine, but also to contextualize it in different parts of their websites, partner’s websites or to promote, and publish that same information on different communication channels or devices.

The Flexibility

What is a flexible platform in that context? A flexible platform is one that can integrate any kind of information sources. Such information sources in the context of Volkswagen can be a series of relational dataset schemas spread around the World, Excel spreadsheets, CSV files, old plain text technical documents about past model of cars, semi-structured documents such as webpages, etc.

A flexible platform is also one that minimally impact (if at all) the data consumers if the data structure changes in the system. This is really important since the World we live in constantly changes. This means that things constantly change and we have to reflect these changes in the data we own and maintain. This is why this point is so important, because we want to minimize the impact of the data structure changes that will happen all the time.

Having the flexibility to constantly adapt your data, while minimally impacting the data consumers of the system, enables you to make quick decision to adapt your strategy in a highly competitive World. This flexibility gives you a clear business advantage.

A flexible platform is also one that let you publish your data the way you want, in the format that is needed. Such a flexible platform has to give you access to an interface that give you access to all the functionalities of the platform without having to care about what happens under the hood.

A flexible system is one that can communicate your information on any kind of communication channels, and to any devices that have access to the Web.

Under the Hood

That next generation platform that Volkswagen is currently developing is partly based on a few of the main pieces of the Open Semantic Framework. These pieces help them to reach their goal by helping them giving the flexibility their platform needs.

The first step they gone thru was to create their Volkswagen Vehicles Ontology that is used to describe all the entities they want to index into their platform. The Web Ontology Language (OWL), along with the Resource Description Framework (RDF) is what gives them the complete flexibility on how they can integrate all the pieces of information they want, in a canonical format.

Then they choose to use structWSF (the structured data web services framework). This piece gives them the flexibility to get a series of web interfaces (web service endpoints) to create, update, manage and query their data. This web service layer enables them to do anything they want with their data, from anywhere on the Web. This is possible because all the functionalities of the framework are exposed as web service endpoints. StructWSF also gives them the possibility to communicate their data in multiple different formats. This makes it the perfect flexible system to feed their information in different contexts, in different communication channels or on different devices.

At Volkswagen, structWSF is used to populate, and keep in sync, their Solr and Triple Store instances. It gives them the time to care about the more important aspects of their platform, and to care about how the data should be synced between the various specialized data management systems.

By using structWSF to manage their data, they are able to reach some objectives to make their platform as flexible as possible:

  • To be able to minimize the impact of data changes to the data consumers
    • Because structWSF uses OWL & RDF to describe all the data it index
  • To be able to manipulate their data from anywhere
    • Because all the functionalities of structWSF are exposed as web service endpoints
  • To be able to communicate the information in different contexts, communication channels and devices
    • Because structWSF has, in its core, is designed to transform all the data it indexes in any other kind of format

The Next Step

One of their longer term goal and objective is to analyze their unstructured and semi-structured textual documents to extract some structure out of them, and to index them into their semantic platform. To do this, they are looking at using Scones, which is the structWSF semantic tagger web service endpoint. Scones will use some subject reference structures such as UMBEL to semantically tag the textual document. Once the document as been processed by Scones, and indexed in structWSF, it can now be re-published in different contexts based on the reference concepts that have been tagged to it. This gives them the flexibility to leverage non-structured sources of data and to re-purpose it in different ways by publishing it in different context and in different systems.

This second system will enable them to leverage the investment they made in the past, by writing all these textual documents, and to re-purpose, and re-contextualizing, them in all kind of different contexts.

Conclusion

I think that TribalDDB and Volkswagen make the good decision for their future. Taking the business decision to develop and maintain a completely new kind of information system is not an easy decision to take. I am not saying that they made the good choice to use our pieces of the stack. The decision goes far beyond this. Such a Semantic Platform challenges everything in an organization: the people that takes the decisions, the people that create and manage the data, the people that develop the system, the people that maintain that system, the consumers of the system, the customers, the partners, etc. This is a big decision; whatever the technology stack you plan to use. I congratulate them for the decision they took.

I strongly believe that this was the right decision to take considering the future opportunities they are creating to themselves.

 

 

Benchmark of PHP’s main String Search Functions

Frederick Giasson
I am currently upgrading the structWSF ontologies related web service endpoints along with the structOntology conStruct module to make them more performing so that we can load ontologies that have thousands of classes and properties (at least up to 30 000 of them).

While testing these new upgrades with them UMBEL ontology, I noticed that much of the time was spent by a few number of stripos() calls located in the loadXML() function of the ProcessorXML.php internal structXML parser. They were used to extract the prefixes in the header of the structXML files, and then to resolve them into the XML file. I was using stripos() instead of strpos() to make the parsing of these structXML files case-insensitive even if XML is case-sensitive itself. However, due to their processing cost, I did change this behaviors by using the strpos() function instead. Here are the main reasons to this change:

  • XML is itself case-sensitive, so don’t try to be too clever
  • These structXML files that are exchanged are mostly internal to structXML
  • Their parsing performances is critical

The Tests

This is a non-scientific post about some experimentation I made related to the various PHP 5.3 string search functions. These tests have been performed on a small Amazon EC2 instance using DBG and PHPeD.

[cc lang=’php’ line_numbers=’true’]
[raw]

[/raw]
[/cc]

The first test uses a text of 138 words. That text get exploded into an array where each value is a word of that text. Then, before each iteration, we randomly select a word that we will search, within the text, using each of the 4 search functions.

Note that in the result images below, each of the line in the left-most column are the ones of the PHP code above.

That first test starts with 10 000 iterations. Here are the results of the first run:


The second test uses the same 138 words, but the test is performed 100 000 times:

As we can see, strpos() and strstr() are clearly faster than their case-insensitive counterparts.

Now, let’s see what is the impact of the size of the text to search. We will now perform the two tests with 10 000 and 100 000 iterations but with a text that has 497 words.

[cc lang=’php’ line_numbers=’true’]
[raw]

[/raw]
[/cc]

That third test starts with 10 000 iterations. Here are the results of the third run:

The fourth test uses the same 497 words, but the test is performed 100 000 times:

As we can see, even if we add more words, the same kind of performances are experienced.

Conclusion

After many runs (I only demonstrated a few here). I think I can affirm that strpos() and strstr() are way faster than their case-insensitive counterparts. However, strpos() seems a little bit faster than strstr(), but it seems to depends of the context, and which random words are being searched for. In any cases, according to PHP’s documentation, we should always use strpos() instead of strstr() because it supposedly use less memory.

There may also be some unknown memory considerations that may affect the code I used to test these functions. In any case, I can affirm that in a real context, where queries are sent to the Ontology: Read web service endpoint that hosts the UMBEL ontology, that strpos() is a way faster than stripos().

What is an Ontology?

Frederick Giasson

An ontology is the definition of a vocabulary, and the rules for combining its terms, used to describe things that needs to be communicated.

This is yet another tentative definition of what is an ontology applied for the semantic web. Before explaining that definition, I would like to continue by stating what I think is the main purpose of an ontology:

An ontology as for main purpose to communicate coherent and consistent information.

Different Kinds of Ontologies

Over the years, I tended to use the word “vocabulary,” along with the word “ontology,” in different blog posts and technical documents. However, the usage of each word may not always have been clear. Is an vocabulary an ontology? Is an ontology a vocabulary? Are these concepts synonymous? There is an important distinction to make: an ontology can be a vocabulary, but an ontology is much more than a simple vocabulary.

Ontologies can describe all kind of well-known knowledge representation structures, some simple, and others much more complex. Here is a small list of some of them:

  • lexicons
  • taxonomies, or
  • higher order knowledge description frameworks

In its most basic usage, an ontology will define a vocabulary. It will simply define the terms (words) that belongs to that vocabulary without saying anything regarding the usage of these words.

Then, an ontology could evolve into a taxonomy by defined hierarchical relationships between the terms that compose the vocabulary.

Finally, it can evolve further to become a higher order knowledge description framework that defines more complex usage rules such as: usage restrictions, all kind of relationships between described entities, etc. New knowledge could also be inferred. It is why I say that an ontology is not strictly a simple vocabulary, but that it powerful knowledge description framework.

Knowledge Base

As we saw above, the main purpose of an ontology is to be able to create a coherent and consistent knowledge base of information that can get communicated. So an ontology is a kind of language that let you create knowledge bases that are consistent, coherent and where new knowledge can be inferred. That is done by following the usage rules defined in the ontology.

However, there is another important aspect to take into account: an ontology will describe knowledge that is coherent and consistent, but according to the own World view of that ontology. This means that two ontologies, describing the same domain of knowledge, could consistently and coherently describe information according to their view of the World.

Let’s take an example. Let’s say that two book stores developed their own ontologies to describe the books they sell. Both companies sell books. There are good chances that they will use the same vocabulary to describe their books. However, the usage rules between these terms may differ between the two book stores. One of the book stores could say that a proceeding is a specialized kind of book. But the other book store could say that no, a proceeding is not a specialized kind of book, but that it is a document just like a book. So, both would describe a proceeding as a document, but one would have different interpretation rules about what a book really is. As you see, both book stores use the same vocabulary to define their library of books, but they interpret their meaning differently. If the two stores would have to exchange information about books in the future, they won’t have many difficulties because they are probably sharing the same vocabulary, but the interpretation of that information may differ. The result of these potential differences in their interpretations may be where a book will be classified into the store; or how their customers could search for a specific book, using different filtering criterias; etc.

This is not different than what happens in our daily lives: is there a day in your life when you don’t hear people arguing about different point of views? It is exactly the same thing that happens here. We potentially all live and see and the exact same events, images, sound, etc.; but we may all have a different interpretation of these things.

Ontologies in the Open Semantic Framework?

Ontologies are so flexible that we choose to make ontologies the “brain” of the Open Semantic Framework.

We wanted to use the most flexible knowledge description framework that would enable us to integrate any possible information sources that have been describe using any existing kind of simple, or really complex, knowledge representation structures such as simple: lexicons, taxonomies, relational schemas, etc. By using ontologies as its central piece, OSF is a flexibly data integration framework that can consolidate information from various, heterogeneous, sources of information.

If we remember the definition we started with, ontologies are not just about describing terms and their relationships in a coherent and consistent way. The ultimate purpose is to communicate that information. It is what the structWSF part of the Open Semantic Framework does: it let any kind of system that have access to the Internet to send, receive and manipulate information in multiple formats from a series of web service endpoints.

More Reading

Finally, I would suggest you to read Mike’s Intrepid Guide to Ontologies to have a better understanding of where ontologies come from, how they works, what other formats exists, what are the different approaches to ontologies and what tools currently exists to work with ontologies.