Geek Answers Handbook

When should you use Semantic Web Technology?

My goal is to implement a semantic search for my data. My data domain contains professional skills. I need additional relationships for my data structure, such as "sameAs" and "wideer". So, for example, if I were looking for a "relationship database", I would also expect "SQL" and "RDBMS" as the results.

My question is whether it makes sense to describe this relationship with RDF (semantic web technology). My skills are already hosted in the MySql database, so I would only create relationships (for example, the same or equal) between skills using semantic web technologies. Is this approach over engineering for my scenario? I do not know when to use semantic web technologies.

+4
source share
3 answers

There are several things to consider:

  • Will your database be a “data warehouse”, that is, autonomous or (ultimately) aimed at interconnecting with the wider world? That is, you find yourself normalizing entities, concepts, etc. that are already available as Linked Data, and / or would you like other people to start linking to your data? For example, do you want to create semantic markup web pages that search engines will search for?
  • You need some kind of conclusion, for example. considering broader as a transitive property, so your application knows that if X is a broader concept for Y and Z is a broader concept for X, Z is also a broader concept for Y.
  • Is there a semantic web solution out of the box that is more or less suitable for you to use? The ability to be based on structure, for example. The Linked Media Framework can be very crucial when it comes to the viability of semantic web technology to use.

If your answers are definitely negative, I would start working on it using "traditional" technology.

If in doubt, you can start development, but keep problems such as stable identifiers deep in your mind. During development, you can try to import it into LMF or write a mapping to D2R and see if it adds something extra to the table.

You can also try browsing / requesting the Stackoverflow equivalent for the semantic Web .

+6
source

If you are more interested in a “wider” / “narrower” type of relationship, perhaps SKOS might be the appropriate approach. It introduces some “soft semantics” (and not my words) into thesauri, classification systems, etc., which seem to be the data you have.

+1
source

Of course, it makes sense to describe these relationships using RDF. Regardless of whether you need semantic web technology, it depends on how you want to use structured data. If you want to be able to explore a large number of related concepts, you can present the data as triples of the RDF and cross the RDF graph accordingly. If you need complex queries that combine several questions, SPARQL may be the answer that again needs the data to be presented in RDF. Also, if you want to automate data management tasks, using Semantic Web technologies would be a good choice. Therefore, basically it all depends on what tasks you want to perform and how.

0
source

More articles:


All Articles