“Tình trạng của Semantic Web”: Ý kiến của tôi (Nguyên văn: “State of the Semantic Web”: My Opinions)

Danny Ayers asked a few questions on the Semantic Web mailing list, and so here are my answers to his questions (his questions are copied word-for-word).

Below are my own personal responses to the questions, and are based upon not only my experience with OpenLink Software but other experience in previous workplaces and academic institutes. They don�脌 necessarily represent the thoughts or beliefs of OpenLink Software, or any of my current or previous colleagues. They are also thoughts which have been quickly cobbled together, so I am sorry for any mistakes.

Obviously Semantic Web technologies potentially have a big role within the corporate Intranet. How are things going there?

I do think that Semantic Web technology is definitely key to a good corporate intranet, as it allows you to interconnect internal resources. Tim Berners-Lee�䏭 project at CERN involved connecting people, locations and projects - and the web grew from there. Linked Data is essential for doing the best that you possibly can do when it comes to corporate work, in other words, you should be able to find something very quickly and very easily and Linked Data offers that. Semantic Web technology, coupled with ODBC and Database Virtualisation is an important area for large corporation - tools such as OpenLink Virtuoso can provide this.

I think businesses are starting to realise the need for interconnected information in the workplace, and there are some organisations who have already adopted a Linked Data approach. At OpenLink Software we “eat our own dog food”, which means that we use OpenLink Software internally (on internal servers and work machines) and externally (on the publicly available web servers), and therefore we are in tune with the Linked Data provided across the business and out into the cloud of Linked Data.

Money! What�䏭 the current status of funding for semweb research in academia? Inside big corps? Gov. orgs? Funding from VCs etc?

At the university that I went to for my undergraduate degree (Oxford Brookes University) they had an interest in Agent Technology, and the Semantic Web was one of their interests. However, funding for research was low and funding was diverted to their strong area which was Machine Learning. At the research institute that I went to for my internship in 2006 (KMI, The Open University), research for Semantic Web and Web 2.0 technologies were reasonable, the research team were stable and they managed to create some very nice applications. At the university that I will be going to for my postgraduate studies (University of Bristol), funding is provided in various areas of Intelligent Systems and this includes knowledge-based topics such as the Semantic Web / Linked Data, the university also has a Web Futures team which is quite active in the Semantic Web community, unfortunately funding only seems to be for PhD Students and Researchers and so I am currently looking for funding for my masters project.

The UK Government also seems particularly interested in Semantic Web / Linked Data technology, and they have adopted

What�䏭 the range of application of RDFa in several areas.

The BBC also are looking into several methods of Linked Data including content negotiation and semantic annotation. The research and development that the UK Government and the BBC are putting into Semantic Web / Linked Data can be used for other projects and further research/development by other organisations.

What�䏭 the range of application of RDF like nowadays? (Obscure examples would be nice)

RDF is all over the place, it only takes firing up the OpenLink Data Explorer and browsing for a little while to find all kinds of information. The interesting thing is that RDF won�脌 only come from explicitly making RDF for your resource, it will also come from sponging/scraping and from transforming other information which may involve some neat Data Mining techniques but will often just involve some simple transformation rules based on web services or POSH (Plain Old Semantic HTML).

What is the significance of recent interest in Semantic Technologies (those without necessarily having any tie to the Web)?

With the recent interest in the portability and the ownership of data, the issue of the links between data is raised. This interest is not only from developers, but also from the average user of the web. The butterfly effect has begun, with more and more articles appearing in the media (including the BBC online, ComputerWeekly Magazine, BCS ITNow Magazine, Communications of the ACM etc etc), soon a lot of people will realise this better way of working.

How far does RDF+ SPARQL (+RDFS) get us? Where might OWL(2) take us? Is there any conflict between these directions?

I see no conflict between any of the technologies: RDF, RDFS, SPARQL or OWL… whichever versions of any of them. I say this because I feel that we should appreciate the overall effect of any kind of addition to the current document-based web. It will be a while before OWL 2 is fully exploited, and this is just because of people understanding the technology.

How have the Linked Data initiatives changed perceptions in the use of RDF?

Linked Data highlights the important parts of the Semantic Web which are ideal for business and academia alike, but particularly business. Interconnected Information across the HTTP web using well described relationships is what organisations need in order to work efficiently, and this is what Linked Data offers. Academia can also benefit, as it provides an interesting method of establishing interdisciplinary knowledge. RDF is at the heart of Linked Data, it is the simplest way because Linked RDF is a graph of Web Data Source Names (WDSNs), where each WDSN can be accessed for a further description. However, RDF as a triple format, is not necessarily the only way, and I am very much open to other suggestions and theories.

How�䏭 the chicken? How�䏭 the egg? (aside - we seem to have a decent supply of data now - but where are the UIs/hooks into existing UIs/never-before-considered applications?)

They are coming. It is a complicated process because the people who tend to be graphics designers and user interface designers have, historically, never realised the full potential of topics such as the Semantic Web and Linked Data. However, this is changing, and probably thanks to Microformats, as they have highlighted the need for some kind of semantics. Keep an eye out for the OpenLink Data Explorer Browser plugin, and I expect that there will be some other rather neat little User Interface systems in the near future as people realise that now that we have data in a common format we can plug all kinds of user interfaces in (hence my blog post on WUPnP a while ago). So they are coming!

Has the role of the W3C changed in this context over the past few years?

I�脎e always liked the W3C because one of my sayings is “well you�羮 never trust a scaffolder if they put up scaffolding in a non-standardised way, and the same should be for web development”. The W3C have always provided the standards, which I prefer to call “suggestions for a good web app”. As for it changing, I don�脌 think it has, although maybe it has got a little more community orientated.

Can we still speak of “The Semantic Web Community” as a (reasonably) unified whole? Should it be?

No and no. The reason I say this is because different people in the “Semantic Web Community” are in it for different reasons, they have different requirements and different interests. In fact, as long as people keep respecting other peoples thoughts then we can actually learn from each others successes and errors, even if they are in different areas and have different requirements. So, I don�脌 think that it should be a unified whole…. that would be like saying that there should only be one denomination of a particular religion, which doesn�脌 make sense because we might as well work together even if we think and feel differently.

Have the attitudes of the developer community at large changed much towards the Semantic Web? (Did SWEO help?)

I feel that SWEO only really helped people who had started to look into the Semantic Web. I think there needs to be a lot of work on people who just dismiss the Semantic Web because either “AI didn�脌 work” or “It�䏭 too academic”, without looking into the subject deeper (if they did then they�羮 realise that its not really AI and its certainly not just for academics). It is a shame that SWEO formally stopped, but hopefully the Semantic Web Community as a whole can help to introduce Semantic Web / Linked Data concepts into even the most distinct management, web development and web user circles.

How has/will blogging influenced the Semantic Web?

Interesting question. I do blog quite a bit about the Semantic Web and Linked Data, and I do wonder how many people really digest some of the things that I suggest. The Semantic Web blogosphere tends to be quite “demystifying” of topics within the Semantic Web, which is great for Evangelism. I also feel that other areas of blogging interest such as Web 2.0, DataPortability and Mobile Software Development have also slightly swayed some of the applications and even some of the standards within the Semantic Web.

As for blogging as an application, there are still some things we can do to blogs to make them more Semantic Web savvy… an example would be completely distributed conversation threads across blogs and forums which is entirely possible using Linked Data, and then queryable using something such as SPARQL. Tagging in blogs is getting better, and will eventually help with hooking into subjects provided on Linked Data systems such as DBpedia, Umbel, Wordnet and MOAT.

Initially the Web 2.0 ‘movement�� had little or nothing to do with the Semantic Web (beyond the lower layers of the stack) - is there any evidence of change there?

Interestingly, if you look at early versions of the Web 2.0 page the “Semantic Web” had been mentioned as being a method of Web 2.0. As I mentioned in the answer to the previous question, blogs and forums, or more socially aware applications assist with some topics surrounding the Semantic Web. Linked Data is data orientated, whereas Web 2.0 about usability and social-ability. The two areas are separate, but not opposite, in fact they can work together… although we can hope to shake off the versioning system, as Web 2.0 is not really any better than anything that we had previously.

Slightly tangential - where do you see social networking going? - supplemental: assume the fad prognosis - what�耱l be next?

Social Networking is dead, the Social Web is now starting to appear (which is completely distributed Social Networking apps coupled with socially-aware object description directories).

Is there yet any compelling, user-friendly application that is solidly based on the Semantic Web (and within that definition I�耱l include linked data and suchlike broad Web connectivity)?

For friendly applications we can look to OpenLink Software for:

OpenLink Data Spaces (ODS) (of which there is an active instance available at: Myopenlink.net)
OpenLink Data Explorer Extension
Other applications to look for which immediately come to mind:

Zitgist Data Viewer
Knowee
bonus: if I want to show Mum how cool the semweb is, without blinding her with triples, where do I start?

Graphs are the same as Objects… Objects are easier to describe to those who are not-so good with computers.

Of the old layer cake, we seem to approaching the point where some of the upper layers don�脌 seem far off being ready for prime time: Rules, Logic, Proof. Too optimistic?

I think they will be coming soon, but it is important that the average Web Developer doesn�脌 go straight for those top level parts, they have to understand Linked Data triples first.

Again with the cake: we know we need Trust - but whatever happened to Signature, Encryption?

I�联 not too clued up on the trust issues and solutions, but I imagine that HTTPS, OAuth and OpenID will be able to help… in addition to reciprocal links and handy inference techniques. PGP may also be useful.

Named graphs are the future?

I thought that Named Graphs were Past, Present and Future. Hash-based URIs highlight this kind of structure, for example:

Named Graph URI: https://example.org/events/socialevents
Object within the Named Graph: https://example.org/events/socialevents#se123
Are we done with new specs yet?

I doubt it, but that isn�脌 a bad thing.

Any impact anticipated from HTML5?

I�联 unsure. I wrote a while back on my blog about how the authors of HTML5 had written incredibly anti-XML and anti-standards comments into the HTML5 (terribly unprofessional!). They have some interesting additions, but I personally don�脌 think that I�耱l be using the HTML5 standard. I think that it will influence Semantic Annotation specs such as XHTML+RDFa and Microformats, and probably has already.

If there was (is?) a Web 2.0 cake, no doubt it would now include OpenID and OAuth - how compatible are these/can these be with the semweb tech we know & love?

As I mentioned in a previous answer, they can definitely be used with other Semantic Web technologies… and have done already. OpenLink Data Spaces also provides OpenID and OAuth servers.

Not unrelated, there�䏭 a fair bit of similarity between OpenID Attribute Exchange and RDF, as well as what appears to be a parallel stack to the (Semantic) Web with XRDS/XRI/XDI etc. Is independent invention of this nature a good thing or not?

Good thing, as I believe that different people have different interests, different skills and different applications and so therefore parallels occur. There aren�脌 many new concepts in the Semantic Web, or the Web in general… it is the combination of these things, the optimisation, the manipulation and the representation that makes a nice app!

There�䏭 always been a Semantic Web roadmap - has its destination changed?

The Semantic Web roadmap hasn�脌 changed. It has, however, given birth the Linked Data, which is a lot more socially and business orientated as it is focused on completely distributed knowledge objects and pulls together from the very old “Data by Reference” idea which has been around since (before) the dawn of Computer Science.

What obstacles are there?

Clearing up misunderstandings and misconceptions of web developers who think that the Semantic Web is “just for academics”.

Event/comm-related things - IM, XMPP, Twitter even - where�䏭 the semweb in all that?

Conversations play a big role in the Universe of Discourse, and the perfect way of describing who-says-what and what about is by using Linked Data. “This Person (their URI) says ‘X, Y, Z�� (URI) on forum A (URI) which is about topic ‘B�� (URI) and ‘C�� (URI)”.

Mobile Semantic Web - how�胩e we doing?

I�联 unsure about the answer to the question as I am not much of a user of the Mobile Web, let alone the Mobile Semantic Web. A nice Mobile Semantic Web application I have encountered though is DBpediaMobile. Which allows you to see a map of the things around you which are fetched from DBpedia.

Ubiquitous Semantic Web - how�胩e we doing?

Similar answer to the answer about the Mobile Semantic Web.

Jim Hendler�䏭 question: where are the agents?

My first answer for this is:
They are here. Human users are agents. Semantic Web Search Engines (such as Swoogle and Syndice) are agents. Semantic Web / Linked Data Viewers are agents.

My second answer for this, and if you are talking about intelligent agents:
They are coming. I think there is quite a bit of work on agents using RDF, but this is in a closed knowledge base. The interesting thing will be when the agents are adapted to feed on the nice Linked Data sets that we have.

What are the best next actions to carry this Grand Project ™ forward?

Huh? Get as many people interested in Linked Data as possible, get academia and business to talk together (Knowledge Transfer projects may be useful).

Loose question - while it doesn�脌 make much sense to say when the Web was/is finished (2 hosts? 2 billion?), but barring disasters, on what kind of timescale do you think we�耱l see a significant qualitative difference in the Web at large due to Semantic Web technologies?

I�联 not entirely sure if I understand your question, because my answer would be that there shouldn�脌 be a qualitative difference between the Web and the Linked Data Web, I imagine that both with be in some kind of symbiotic relationship with each other.

Anyone noticed any serendipity recently?

Errrr:

Serendipity [ https://en.wikipedia.org/wiki/Serendipity ]… nope
Serendipity [ https://en.wikipedia.org/wiki/Serendipity_(weblog_software) ]… nope, not used it
Serendipity [ https://en.wikipedia.org/wiki/Serendipity_(book_series) ]… nope, not read it
Serendipity [ https://en.wikipedia.org/wiki/Serendipity_(film) ]…. ooh, Kate Beckinsale!(… she was awesome in the Underworld [ https://en.wikipedia.org/wiki/Underworld_(series) ] series.). I don�脌 think I�脎e seen the Serendipity film though.

Trả lời các câu hỏi của Danny Ayers về Semantic Web
Danny Ayers đã đặt một số câu hỏi trên mailing list của Semantic Web, và dưới đây là những câu trả lời của tôi (các câu hỏi được giữ nguyên).

Những câu trả lời này dựa trên trải nghiệm cá nhân của tôi, không chỉ từ công việc tại OpenLink Software, mà còn từ kinh nghiệm làm việc trước đây và các tổ chức học thuật mà tôi từng tham gia. Chúng không nhất thiết phản ánh quan điểm chính thức của OpenLink Software hoặc bất kỳ đồng nghiệp nào của tôi, và vì được viết nhanh nên có thể có sai sót – mong các bạn thông cảm!

1. Semantic Web có vai trò gì trong Intranet của doanh nghiệp?
Tôi tin rằng Semantic Web đóng vai trò rất quan trọng trong việc xây dựng một hệ thống Intranet hiệu quả trong doanh nghiệp. Nó giúp liên kết dữ liệu nội bộ giữa con người, địa điểm, dự án và các tài nguyên khác.

Tim Berners-Lee đã bắt đầu với ý tưởng kết nối thông tin tại CERN, và từ đó Web phát triển mạnh mẽ. Linked Data chính là chìa khóa giúp doanh nghiệp tận dụng tối đa dữ liệu của họ – nghĩa là bạn có thể tìm thấy thông tin một cách nhanh chóng và chính xác hơn.

Khi Semantic Web kết hợp với ODBC và ảo hóa cơ sở dữ liệu, nó trở thành một công cụ mạnh mẽ cho các tập đoàn lớn. Các nền tảng như OpenLink Virtuoso giúp hiện thực hóa điều này.

Nhiều doanh nghiệp đã nhận ra tầm quan trọng của thông tin liên kết trong nội bộ và một số công ty đã áp dụng mô hình Linked Data. Ở OpenLink Software, chúng tôi áp dụng phương châm “eat our own dog food”, tức là tự sử dụng công nghệ của chính mình trong cả hệ thống nội bộ lẫn các máy chủ công khai. Điều này giúp dữ liệu của công ty dễ dàng hòa nhập vào mạng dữ liệu mở trên Web.

2. Tình hình tài trợ cho nghiên cứu Semantic Web hiện tại ra sao?
Trong học thuật:

Khi tôi học tại Oxford Brookes University, họ quan tâm đến Agent Technology, nhưng ngân sách cho Semantic Web khá hạn chế, thay vào đó, họ tập trung vào Machine Learning.
Khi tôi thực tập tại KMI, The Open University (2006), nguồn tài trợ cho nghiên cứu Semantic Web khá ổn định, giúp họ tạo ra nhiều ứng dụng ấn tượng.
Tại University of Bristol, tài trợ được phân bổ cho Intelligent Systems, bao gồm Semantic Web / Linked Data. Tuy nhiên, phần lớn ngân sách lại dành cho nghiên cứu sinh Tiến sĩ và nhà nghiên cứu, khiến tôi phải tự tìm nguồn tài trợ cho dự án Thạc sĩ của mình.
Trong chính phủ:

Chính phủ Anh tỏ ra khá quan tâm đến Semantic Web / Linked Data, và họ đã bắt đầu triển khai công nghệ này trong các dự án công.
Trong doanh nghiệp lớn:

Các công ty như BBC đang thử nghiệm Linked Data với nhiều phương pháp khác nhau, bao gồm content negotiation và semantic annotation. Những nghiên cứu của BBC và chính phủ Anh có thể giúp mở ra cơ hội ứng dụng rộng rãi hơn.
3. Ứng dụng của RDFa trong thực tế?
Hiện nay RDFa có nhiều ứng dụng thú vị, đặc biệt trong lĩnh vực xuất bản nội dung trên web, giúp công cụ tìm kiếm hiểu rõ hơn về dữ liệu. Ngoài ra, nó còn được sử dụng trong hệ thống tri thức và thương mại điện tử để mô tả sản phẩm, dịch vụ một cách rõ ràng hơn.

4. Ứng dụng của RDF hiện nay ra sao? Có ví dụ nào đặc biệt không?
RDF xuất hiện khắp nơi, bạn chỉ cần mở OpenLink Data Explorer để khám phá hàng loạt nguồn dữ liệu. Một điểm thú vị là RDF không chỉ đến từ việc tạo RDF một cách trực tiếp, mà còn đến từ sponging/scraping và chuyển đổi dữ liệu có sẵn bằng các kỹ thuật khai thác dữ liệu (Data Mining).

Ví dụ đặc biệt:

BBC sử dụng RDF để liên kết thông tin về chương trình truyền hình, diễn viên, phim ảnh.
DBpedia trích xuất dữ liệu từ Wikipedia để tạo thành mạng dữ liệu mở.
Linked Open Data Cloud chứa hàng trăm tập dữ liệu được liên kết với nhau.
5. Tại sao công nghệ Semantic lại nhận được sự quan tâm gần đây?
Ngày càng có nhiều bài viết trên các tạp chí công nghệ như BBC Online, ComputerWeekly, Communications of the ACM, v.v. Điều này cho thấy nhận thức về Linked Data và quyền sở hữu dữ liệu đang lan rộng không chỉ trong giới phát triển mà còn trong cả người dùng phổ thông.

6. RDF + SPARQL đưa chúng ta đi xa đến đâu? OWL 2 có vai trò gì?
Không có sự xung đột nào giữa RDF, RDFS, SPARQL hay OWL. Mỗi công nghệ này bổ sung cho nhau để mở rộng khả năng của Semantic Web. Tuy nhiên, OWL 2 sẽ cần thời gian để được khai thác toàn diện vì đây là công nghệ khá phức tạp.

7. Sáng kiến Linked Data đã thay đổi cách sử dụng RDF như thế nào?
Linked Data giúp Semantic Web trở nên hữu ích hơn đối với doanh nghiệp, vì nó liên kết thông tin một cách có hệ thống thông qua HTTP. Các tổ chức có thể tận dụng điều này để tối ưu hóa quy trình làm việc và chia sẻ dữ liệu hiệu quả hơn.

8. Giao diện người dùng cho Semantic Web đang ở đâu?
Hiện nay có một số ứng dụng UI hữu ích như:

OpenLink Data Spaces (ODS) – Hệ thống mạng xã hội Semantic Web
OpenLink Data Explorer – Trình duyệt dữ liệu Linked Data
Zitgist Data Viewer – Công cụ trực quan hóa dữ liệu Linked Data
Những công cụ này giúp người dùng tiếp cận dữ liệu ngữ nghĩa mà không cần phải biết về triples hay SPARQL.

9. Semantic Web và Web 2.0 có giao thoa gì không?
Dù trước đây Web 2.0 và Semantic Web có vẻ không liên quan, nhưng thực tế Linked Data và Web 2.0 có thể bổ trợ cho nhau.

Web 2.0 tập trung vào trải nghiệm người dùng và mạng xã hội.
Linked Data tập trung vào kết nối dữ liệu theo cách có ngữ nghĩa.
Khi kết hợp cả hai, chúng ta có thể tạo ra những ứng dụng thông minh hơn, trực quan hơn.

10. Tương lai của Semantic Web sẽ ra sao?
Semantic Web vẫn đang phát triển mạnh mẽ, đặc biệt là với sự mở rộng của Linked Data. Một số hướng đi quan trọng bao gồm:

Ứng dụng trong trí tuệ nhân tạo và trợ lý ảo.
Tích hợp với OpenID, OAuth để xác thực dữ liệu.
Phát triển UI thân thiện hơn cho người dùng phổ thông.
Vẫn còn một số rào cản, đặc biệt là nhận thức của cộng đồng lập trình viên, nhưng với sự phát triển liên tục, tôi tin rằng Semantic Web sẽ dần trở thành một phần quan trọng của Web hiện đại.