Some discussion points on wider understandings on semantics in the context of `context` within Fediverse environments

@Ryuno:matrix.org

Here's a question I was always afraid to ask: what is meant with semantics in CS?
I understand syntax and tokenizing. But „adding meaning to words“ isn't that easy to map onto code for me.
Like, @context is still a property on a JSON object with a string as value. At which point is the meaning added?

@stevebate:matrix.org

The json-ld @context defines terms for the purpose of serialization between RDF and JSON, but not the meaning of those terms. The meaning of the terms is defined elsewhere. For AP/AS2, the normative meaning is in the natural language in the W3C Recommendation documents. There's a formal ontology for AS2, but it hasn't been maintained in years.

...

... json-ld does not provide it. You wouldn’t be able to successfully implement an AP application without understanding the meaning of many aspects of the protocol.

@circlebuilder:matrix.org

... I think a problem that semantic web / linked data has always had, is that it stopped short on saying "machine-readable data that conveys more semantic meaning is good" without going the extra mile in selling the vision of how people would actually experience a digital world of linked data. I.e. the focus was on going to a next / "better" technical level, introducing much necessary complexity along the way, without selling the vision on why all this effort this would be worthwhile. This left many people "not getting it" unwilling to adopt.

...

The "machine-readable" of semantic web and the ability to "understand" the data by machines, has been described confusingly many times and what it enables blown out of proportion (if the whole web were linked data we would be able to build apps differently, but it still needs be explicitly implemented). It basically consists of adding more metadata that relate to real-world concepts. Compare it to html with only <div> making all the layout, versus 'semantic html'. It is still actual code and logic that should do something with that.

...

I think when considering linked data there's like multiple 'adoption levels' (though not recognized). You have using well-known ontologies and otherwise writing your extensions using best-practices. This adds a whole layer of useful metadata, and isn't too complext. And then there's going onwards into modeling universal human knowledge, where everything ties together. In this "open world" semantic web the complexity is humongous, as there's no such thing as "universal meaning". Everything depends on context and relationships and can be expressed and sliced in different ways.

@damon/:matrix.org

... I believe CB has a good point in the general view of opinions regarding the Semantic web and the complexities versus benefits and if the work to

1. study and understand it
2. to implement it is worth it

@stevebate:matrix.org

For AP/AS2, two justifications are extensibility (managing property key namespaces) and re-use of existing vocabularies/ontologies (vcard, schema.org, foaf, doap, and many, many more). There are other benefits for those doing RDF/graph work, but I won't include those as justifications.