Welcome to Talis Platform News

Welcome to the latest issue of Talis Platform News.

In this issue we have a further installment of Ian Davis' regular column, Danny Ayers introduces us to Laurian Gridinoc and the work he's doing with the Talis Platform, and we take a look at growing interest in the Semantic Web at large on the back of the recent Web 2.0 Summit and Tim Berners-Lee's pronouncements on the Giant Global Graph.

I hope you find this issue of interest, and welcome suggestions for items you'd like to see covered in future issues. Don't forget to subscribe in order to receive an email reminder and highlights each month, and feel free to join us in IRC conversation over on the #talis channel hosted on freenode.net.

Paul Miller, Editor

Web 2.0 and the Semantic Web - a meeting of minds

As mentioned in the previous issue of Talis Platform News, this year's Web 2.0 Summit in San Francisco included a panel session on the Semantic Web. Interest was whipped up in the hours before the session, with Talis Platform Advisory Group member, podcast victim, and Radar Networks CEO Nova Spivack unveiling their new application; Twine. Twine was described as;

"a new service for sharing, organizing and finding information with people you trust. Use Twine to better leverage and contribute to the collective intelligence of your friends, colleagues, groups and teams. Twine ties it all together."

Twine is still inaccessible to those not in their testing programme, but its announcement was covered with great interest across the Web with luminaries such as Nick Carr, Richard MacManus and Tim O'Reilly being quick off the mark.

Returning to the panel session, video for which is available here, Tim O'Reilly moderated a discussion with panelists;

• Talis Platform Advisory Group member, podcast subject and Radar Networks CEO Nova Spivack;
• Danny Hillis of Metaweb (boss of Talis Platform Advisory Group member and podcast subject Jamie Taylor);
• Barney Pell of Powerset, Inc.

The session was live-blogged here, with some post-event contextualising thoughts here.

Since the conference, there has been a continuing buzz around semantic applications; a buzz brought to a head in a recent post on Read/WriteWeb by Richard MacManus. His 10 Semantic Apps to Watch includes all three of the Web 2.0 panelists... and the Talis Platform.

Notes from Behind the Curtain

Last time I wrote about the benefits of running a shared platform: the possibility of small improvements to benefit all platform users automatically. The other side of that coin is that every mistake we make can impact a lot of users, and could potentially cause a lot of damage. This month I thought I'd take a peek at our strategies for reducing this risk, drawn from our experience with agile development.

One strategy is to release our software frequently. Currently we make a new release of the Platform every month, even if we have very few features or fixes in the release. Any software developer will tell you of the tense moments that accompany putting any system live and the first time is always the worst. However, if you force yourself to do it again a month later, then again and again, you face up to your fears pretty quickly and can start to make real risk assessments based on the evidence of past releases. You also learn what not to do very quickly and those memories are still close to you when you make the next release.

Another strategy we use is to test everything that is humanly possible to test and we're very strict about automating those tests too. We cover our code with several thousand small unit tests and Sam and the other developers run them about 100 times a day, basically every time they add any new code to the Platform. At the same time we write bigger acceptance tests that test how all the Platform components interact. These too are automated so it's possible for a developer to click a button to build and launch a fully working version of the Platform on their computer and run all the tests against it.

Once a developer is happy with a particular feature or fix and, crucially, all the tests can be shown to be passing for the new code, it goes through a peer code review stage. Here we pass a second pair of eyes over the code and check for areas the original developer might not have thought to test. This process is friendly but strict since we want to keep the quality of the Platform code as high as possible at all times. The next stage is a manual acceptance test where we deploy the development version of the platform to a test environment and someone physically confirms that the behaviour is as expected and that we adhere to any pertinent standards. Finally, just before the monthly release we deploy the Platform to our staging area. Here we run 24 hours of performance tests before handing over for the final set of regression tests. We invite application developers to point their applications at the Platform in the staging area and look for any unexpected changes.

As you can see our process to get from drawing board to being deployed live in the Platform is very long and involved. Yet we manage to do this and sustain it while still making a release of the Platform every month. We couldn't do it without our automation practices which make it possible to build, deploy and test the Platform multiple times a day. So far it has worked well for us - we have caught and fixed performance and regression errors before they made it out into the live environment. So far so good, but we need to keep vigilant and do all we can to keep the Platform a safe and trustworthy environment.

A Plethora of Podcasts

We've been producing a series of Semantic Web-inclined podcasts for the last twelve months or so and this is now building into quite a collection, some highlights of which featured on ZDNet recently.

Recent additions to the set include Yihong Ding, Kaila Colbin & Branton Kenton-Dau of VortexDNA, Daniel Lewis (now) of OpenLink, Alberto Reggiori of Asemantics, Peter Bloodsworth of the University of the West of England, and Wordmap founder Bill Hutchison.

Also recorded in the past week, and in various stages of readiness for release, are conversations with Kjetil Kjernsmo and David Norheim, Eyal Oren, the team behind GroupMe!, Inigo Surguy, Mathieu Daquin and Andreas Harth. All of those should become available over the next week or two, and as usual they'll be blogged on Nodalities as they are published, and then added to the catalogue.

We're always looking for new topics, so if you have a topic to discuss or someone you'd really like to hear please do get in touch.

Tim Berners-Lee draws the Giant Global Graph

Tim Berners-Lee raised the profile of the Semantic Web once again, with a blog post in which he aligned semantic technologies squarely with the relationships and connections so key to the current enthusiasm for social networks and the 'social graph.' Tim's talk of the 'Giant Global Graph' provoked a variety of reactions, a number of which were compared here.

We didn't see something new created by Tim's post. What we saw was a restating of some principles at the heart of the Semantic Web, a recognition that the social graph so frequently mentioned in relation to the big Social Networking sites shares many of those principles. Finally, we saw the beginning of an informed discussion that might - finally - see the fruits of many years of Semantic Web research and development surfaced in language that can be used in conversation with the pragmatists building the mainstream Web of today, aligned to technologies and techniques fitting for that Web, rather than simply making the gloomy shadows a bit more pronounced.

The discussion continues.

Talis Platform Early Access Programme

Danny Ayers talks with Laurian Gridinoc, a PhD student at the Open University's Knowledge Media Institute (KMI).

The topic of my research is semantic browsing, which is about how to bring semantic information within a regular web browsing session — to actually display a 'semantic layer' upon a particular resource. This layer provides semantic links (in addition to the existing ‘syntactic’ ones) that are coherent with a particular perspective on the data.

I believe in using the semantic web as ‘background knowledge’—using a mesh of ontologies to yield interesting and often unanticipated connections.

Previously to my PhD, I did a Masters in NLP and 8 years ago I co-founded Grapefruit, which is currently one of the leading Romanian branding and interactive design agencies. Oddly, I hold a Bachelors degree in medicine and surgery.

Past projects include Watson, an ontology search engine, where I developed the infrastructure (storage, crawling and indexing.)

I'm currently working on PowerMagpie, which aims to bring the aforementioned semantic layer to a regular browsing session.

On the Talis Platform I'm trying to create a blog engine. The first step is to achieve a purl.org-like URL indirection functionality.

In the future I would like to use NetKernel in order to develop interesting apps over the Talis Platform.

Keep an eye on the wiki for the Talis Platform Early Access Programme, [n2].

To receive notification of new issues of the Talis Platform News please remember to SUBSCRIBE.