Defin­i­tion of semantic web

The semantic web refers to the next stage in the de­vel­op­ment of the world wide web. In what is known as Web 3.0, in­form­a­tion is no longer just linked, but web content is enriched and linked with machine-readable, semantic metadata. The aim is to optimise the in­form­a­tion exchange on the web by enabling machines to dis­tin­guish and spe­cific­ally process machine-readable meanings, i.e. semantic content.

The term 'semantic web' is one of many terms used to define a semantic de­vel­op­ment of the world wide web. In addition to semantic web, the following terms for the global, se­mantic­ally linked in­form­a­tion network are also being discussed:

• Web 3.0: Has been cir­cu­lated by US journ­al­ist John Markoff to describe how machine-readable meanings are being added to the in­ter­act­ive, col­lab­or­at­ive Web 2.0.
• GGG (Giant Global Graph): Used by Tim Berners-Lee, inventor of the www, as a de­scrip­tion of a global in­form­a­tion structure that uses semantic struc­tur­ing of metadata and content; GGG overlaps con­cep­tu­ally with web semantics.
• Linked Open Data: Coined in 2007 to emphasise metadata standards, query routines, and networked semantic data as the found­a­tion of the semantic web.
• Web of data: Defin­i­tion in­tro­duced by the W3C, the World Wide Web Con­sor­ti­um, in 2013 to combine the syntactic and semantic in­ter­con­nec­ted­ness of data in one term.

Until now, the www has been primarily ori­ent­ated toward the syntax of in­form­a­tion. Here, computer programs use al­gorithms that analyse data indexes, keywords, and search queries. Depending on how unique a query is, search engines deliver more or less ap­pro­pri­ate search results (SERP). However, it is important for users and companies that programs process search and user intent as ef­fi­ciently as possible. The semantic web not only aligns with search terms and syntax, but also with meaning values. In this way, machines can find content and un­der­stand and dis­tin­guish their meaning.

For example, if users search for the phrase 'When did Barack Obama’s pres­id­ency begin?', search engines would not simply return 'January 20, 2009', but rather the most ap­pro­pri­ate hits possible for Barack Obama. In the semantic web, machines un­der­stand not only the content but also the meaning of a search query and provide an exact answer. Moreover, the analysis of meanings in the semantic web includes not only text, but also images, sound, numbers, and symbols – in other words, all features that carry meaning.

Entities and on­to­lo­gies are among the core com­pon­ents of the semantic web. 'Entity' is a term from semantics – it consists of an iden­ti­fi­er and as­so­ci­ated at­trib­utes. As an example, 'Barack Obama' would be the iden­ti­fi­er in an entity, while in­form­a­tion such as 'US Pres­id­ent', 'lawyer', 'demo­crat' are the at­trib­utes, i.e. de­script­ive prop­er­ties. Entities, in turn, can be related to one another and them­at­ic­ally related or different.

If entities stand in a context to one another, they’re called 'on­to­lo­gies'. On­to­lo­gies are ordered sets of in­form­a­tion and logical state­ments that are for­mu­lated in a way that is readable for humans or machines and that establish con­nec­tions and show re­la­tion­ships.

Entities and on­to­lo­gies are essential for the semantic web. Programs use them to un­der­stand re­la­tion­ships between words, sentences, images, and char­ac­ters, in­tel­li­gently filter multiple meanings and duplicate content, interpret web content, and them­at­ic­ally dis­tin­guish entities. In this way, a rich knowledge network is created that consists not only of un­struc­tured in­form­a­tion, but also of keywords and addresses. In the future, ar­ti­fi­cial in­tel­li­gence will be able to su­per­fi­cially search the ac­cu­mu­lated knowledge of the www, and un­der­stand and interpret it in a more goal-ori­ent­ated manner.

To realise the semantic web, computer programs must learn to extract meaning. This is only possible if existing or new www content contains struc­tured data that is for­mu­lated in a machine-readable way. Struc­tured data is for­mu­lated using specific standards and clas­si­fic­a­tions and is encoded on websites in the form of a schema mark-up and in-page mark-up. Struc­tured data allows programs to clearly dis­tin­guish, for example, 'bank' as a financial in­sti­tu­tion from the object 'bank' referring to the sides of a river. In turn, a uniform machine-readable language requires Semantic Web Standards, as for­mu­lated by the W3 Con­sor­ti­um.

Other ap­proaches to uniform semantic web standards include the Con­tex­tu­al Browsing Language (CBL), which describes re­la­tion­ships between in­form­a­tion, and the Web Ontology Language (OWL), which organises and clas­si­fies in­form­a­tion hier­arch­ic­ally. In addition, the following mark-ups and standards, among others, help create semantic meta-state­ments, standards, and rules:

• RDF/RDFa (Resource De­scrip­tion Network in At­trib­utes): Used to describe websites in detail to make logical, semantic state­ments about arbitrary content, and can be extended by RDFa to integrate RDF with XML.
• URI (Uniform Resource Iden­ti­fi­er): Iden­ti­fies in­form­a­tion units and points to available Linked Open Data (LOD), i.e. con­tinu­ing data in HTTP documents.
• RIF (Rule In­ter­change Format): Defines rules according to which con­tex­tu­al meaning is created.
• Dublin Core: A standard for metadata embedded in digital documents and for machine-readable in­ter­pret­a­tion of elements for­mu­lated in RDF.
• RDFS (Resource De­scrip­tion Framework Scheme): Iden­ti­fies the RDF vocab­u­lary and specifies the structure and syntax to be used.
• SPARQL (SPARQL Protocol And RDF Query Language): Serves as a query language and protocol for content from the RDF system, which consists of logical de­scrip­tions and re­la­tion­ships of data.

The ad­vant­ages of the semantic web should not be un­der­es­tim­ated. Companies are already relying on it to adapt to the di­git­al­isa­tion of the business world. Those who analyze pur­chas­ing and search be­ha­viours of customers and target groups can provide per­son­al­ised in­form­a­tion and generate more traffic. In online marketing, ad­vert­ising that is geared to the semantics of web content can be better adapted and linked to keywords that cor­res­pond to a company's services and products.

For search engine optimised websites, too, it's not just a matter of good keywords, but of semantic in­form­a­tion that struc­tures content and ensures a machine-readable in­form­a­tion ar­chi­tec­ture. Be sure to include struc­tured data in websites and make web content as mean­ing­ful as possible using semantic standards. In this way, you can improve your search engine ranking and can be found by the target groups you wish to attract.

The semantic web is still in its infancy, but the first steps towards its real­isa­tion have already been taken. For example, the pos­sib­il­it­ies of the semantic web can be seen in Google's Rank Brain, which can them­at­ic­ally assign search queries pre­vi­ously unknown to the algorithm. Google's image search already 're­cog­nises' what users are searching for and delivers them­at­ic­ally similar image results. Similarly, Google's Knowledge Graph feature is able to recognise semantic entities and display the most important related or connected in­form­a­tion in addition to search results. Similarly, Google's Rich Snippets and rich cards prepare struc­tured data in the form of in­form­a­tion carousels and excerpts from websites.