What are the key milestones for Semantic Web?

It would be really useful to understand what people perceive to be the key milestones for Semantic Web.

Here are the most useful milestones I can identify:

Milestone 1: Begin.
Milestone k: Something better than Milestone k-1
for 1<k<∞

The limit as k → ∞ might be something useful.

(Inspired by Gall's law.)

EDIT: In hindsight, apologies... I guess the above answer might be a little facetious for what is an interesting question. I also think that this is perhaps the hardest question that I've seen on this site (I haven't seen "What is the Semantic Web?" asked yet...)

I feel like I owe a reasonable effort at an answer, and I actually tried to define some milestones, but I wasn't buying my own answer, so I'm not going to post it (:P). Being honest with myself, I think that Semantic Webby stuff will slowly leach into the mainstream, primarily through lightweight adoption by big players... In any case, I can at least give my version of why I think the question is hard, and what I meant by the above "answer".


The point I was probably failing to make above was that milestones are inherently difficult to identify in hindsight, let alone to foresee. The Semantic Web is a nebulous, complex thing... you could draw analogy to the current Web. Even in hindsight, what milestones would you identify for the Web? How would you connect together the milestones into a coherent picture? Methinks that my picture and your picture and Fred's picture would look entirely different. (...things that one might naturally identify as milestones were often long, gradual, hard-fought battles of attrition, which eventually hit critical mass and flourished.)

The Web's inception was with little fanfare. It was a simple system that worked and scaled, and which evolved over time into a complex system that worked and scaled... it evolved from small steps, each forward step making incremental improvements... cul-de-sacs were also well explored: an important aspect of progress.

Looking towards the future of the Semantic Web, how can one go about meaningfully plotting milestones between A and B when we have no idea where B is, what it looks like, or even what direction its in? Will we even recognise it when we see it? All we know is that B isn't where we are now, so we gotta keep moving, follow our respective hunches, and keep others up-to-date on what we find. Any course that anyone tries to plot in advance will not only be subjective, but almost certainly naïve.

<aside>...at this juncture, I can't resist pointing to the ridiculous quagmire that is the middle/latter parts of the Semantic Web Wikipedia article, and all of the empty, groundless and ultimately pointless speculation that it contains. ...it's the first hit in Google and it certainly doesn't help our cause. (However, I commend the brave souls who work to improve the article.)</aside>

In my opinion, we should fixate less about the future (it might never happen) and should focus on improving what we have now, and keep doing that. ...be flexible, adapt, and just keep making stuff better than it was before. We should downplay stacks, layercakes and milestones (which ultimately act as sticks to beat our incremental successes with). Forget about the macro (over which we have no direct control) and focus on the micro (over which we have control). ...leave milestones to the historians.


EDIT2: As per William's comment. After this—you'll be glad to hear—I'm taking a sabbatical from long-ass answers on this site. I promise!

Here's the Milestones I was playing around with, but I don't entirely buy...

...very much a Linked Data perspective, and very much subjective.


Milestone 1.1: Get Linked Data to be the de facto consensus—the obvious solution—for published legacy, openly-available structured data... get those who can to open up their databases/structured corpora... the LOD project and RDB2RDF WG have made a start.

Milestone 1.2: Integration of Linked Data exporters into mainstream Web publishing tools, so that the broader Web is churning out (high-quality) RDF as a "background process"... reach them where they are... efforts like Drupal and identi.ca are a start.

Milestone 1.3: Adoption for techniques/infrastructures which automatically create and maintain high-quality links between published datasets... efforts like Silk and DSNotify head in that direction.

Milestone 1.4: Refinement of the Linked Data best-practices and literature. Refinement of notions of Linked Data quality. Better educational material. Avoid unnecessary hype. Simplify current standards or promote grass-roots subsets of the (often prolix) W3C standards. Deprecate, deprecate, deprecate... fingers crossed for the RDF WG.

Milestone 1.5: Sustainable architecture/methodologies for creating and maintaining vocabularies. We're going to need ad-hoc, well mapped, well modelled, and well agreed-upon vocabularies for the various domains.

Milestone 1.6: User involvement... create (preferably decentralised) user feedback-loops: get users to generate and curate Linked Data corpora using the usual sociological/behavioural "tricks". Offer tangible and immediate rewards for tangible efforts.


Milestone 1.x: Lots of high-quality, well-interlinked, structured data, effortlessly managed and published on the Web, ready to be consumed.



Milestone 2.1:Sustainable, accessible, centralised warehouses of Linked Data... centralised warehouses are probably needed to bootstrap adoption and complement other developments. Create APIs to really help publishers... e.g., find URIs to map to, find classes/properties to use. Develop appealing simple use-cases for (i) developers, (ii) normal users. Do usability studies. Don't be lazy regarding data quality... Sindice et al. heads in the right direction.

Milestone 2.2: Working decentralised architectures and techniques. Get federated SPARQL querying working. Get live query-processing and browsing of Linked Data working... SPARQL 1.1, tabulator in that area.

Milestone 2.3: Creative, appealing, usable applications and mash-ups.


Milestone 2.x-3: Linked Data applications that people use without being told to.

Milestone 2.x-2: Linked Data applications that make people's lives easier.

Milestone 2.x-1: Linked Data applications that people (think they) can't live without.

Milestone 2.x: People wondering how they ever managed without Linked Data.



Milestone 3.1: Business models for running Linked Data services and hosting Linked Data. Seek out sustainable revenue streams to pay for hosting... advertising, pay-per-click, pay-per-triple, directing traffic, donations... Limited revenue = limited hardware = limited services = transient services and data exports... we're far behind on this.

Milestone 3.2: Have legal issues sorted. Look at licencing issues. Create recommendations on how to avoid being sued.

Milestone 3.3: Keep research on-track. Investment in timely, high-impact research. Focus on motivation, realistic evaluation, reproducibility of results, portability of techniques. Keep small threads of "blue sky research".


Milestone 3.x-1: The first Linked Data millionaire.

Milestone 3.x: The first Linked Data fortune 500.


The main purpose of Semantic Web (or Web 3.0, for me these two are interchangeable), is to give the machines the hability not only of exchange data, or bits of raw information, but knowledge.

We get that goal by teaching the machines to understand the meaning of the objects that travel across the web. Because of that, there have been development on ontology languages, ontology processors (reasoners are called?), The model RDF (Resource Description Framework), etc, etc. All of these technologies are useful on pursuing the goal of making the machines work with meaning.

One of the goals to reach is to build more applications implementing these technologies in order to make the Semantic Web nearer to the public. And IMHO, the more inventions we do, the better we get at it, and the more mistakes we find to correct: you fall, you learn.

That were my two cents. I'm very new in this world, but I fell in love. Greetings.-

I guess, the technologies of the common, layered Semantic Web technology stack plus propagating RDF as a data model (knowledge representation structure) independent of concrete serializations.

RDF 2 ∨ Linked Data Protocol and API ∨ RDF Data Bus used as defacto for Internet Of Things ∨ Realisation of the Read-Write Web of Data.

Coupled with some design trade-offs and culling of features in order to handle the common use cases of the wild web.