Laptops: We used to like Macbook Pros, mostly becase they run OS X, but really they have a very poor build quality – one ours has started falling apart with a hardware fault that should never have happened (internal frame for the screen broke without any maltreatment).

For build quality our hearts lie with Lenovo Thinkpads, and the X series is particularly good for lightweight portability. Mark’s six year old x40 is still great, and more recently a bit of nosing around on ebay has supplied an x60 and an X200 that are both destined to loose Vista and gain Ubuntu desktop.

Desktops: We have a couple of desktop format machines, although one of them doubles as a demo server, and the other is permanently running server software. Our mantra for desktop format machines is small and fast. Ours are built around Intel processors (dual core i5 and quad core i7 chips). We did flirt with the idea of six core AMD Phenoms, but their performance was less than somewhat-equivalently priced  i7 processors.

Our smallest desktop uses a mini-ITX format and contains a  Zotac H55-ITX mother board (see this arbitrary review). The motherboard takes 1156 pin Intel CPUs with integrated graphics which is handy when we take the machine out as a demo server.  In the office it runs Ubuntu desktop we fill its single expansion slot with a large and fast Nvidia graphics card that gives us two DVI signals to drive two of our favourite IPS monitors for a rock solid display. Finding a nice case for this machine was a bit of a sweat; we ended up with a Silverstone mini-ITX case with replacement fan from silentPC.com.

Our second desktop sized machine runs a Ubuntu server install and the Manchester PLE all the time, and is about to be pulled apart to become a rack-mount server, but for all of a few days it still exists in desktop form. It has a slightly larger micro-ATX format motherboard, which gives us more expansion slots than the miserly single expansion slot offered by a mini-ITX mobo. We’re using an Asus GENE Rampage II motherboard (arbitrary review), a quad-core i7 processor, and 6G of memory that occupies half the available memory slots. For a while we overclocked this machine and stuffed a larger Silverstone case full of yet more silentPC sourced fans to keep the beast cool, but the radical overclocking we were trying was more hassle than it was worth, and we are back to close to stock speeds.

There are a couple of Shuttles knocking around from pre-Hedtek days, but really the joy of Shuttles has waned. They’e cool if you want to start with a barebones machine, but not that much fun to build compared to other alternatives.

Mini-tower servers: Well what’s cheap currently at Insight UK? Actually, we have an HP ML110 G5 that has stood us in good stead running Redmine and a few other useful bits of software for a while, with a bit of extra memory and a replacement CPU, a quad-core Intel Q6600. And we are about to buy a couple of the G6 versions for various uses, Dave to run a work server at home, and Mark to dedicate to our processing our accounts. IBM do a nice line in mini-tower servers, but they tend not to be as cost effective as the HPs unless Insight is doing one of its occasional deals.

Servers: Supermicro, Supermicro, Supermicro. Did I mention Supermicro? We have a few second-hand Supermicros, and have speced and ordered a big fat fast one with dual quad core processors and 76GB of memory for the fishDelish project, where it will run triple stores, execute SPARQL queries and do a bit of web page construction via our Ruby-on-Rails systems. Actually we don’t and won’t own this server, but we are doing all the implementation that is going to happen on it.

And we’ve picked up a few Chenbro rackmount cases with eight hotswap drive bays each that we will use with the above-mentioned Asus motherboards to run the Manchester PLE. When we build these things its gonna be with hardware raid and quality power supplies; the only way to go. Actually, one of us is itching to use something like these Chenbro cased machines to start to play with dbdr to build a high-availability database server.

Blade servers:  ebay is big in our lives. We have three Hewlett Packard BL-class blade server chassis and a variety of BL-25 and BL-35 blades bought on ebay  for the very miserly sums of money. Only one of these blade servers is in use, populated with three blades that we are using to run the infrastructure for our agile development course, to wit, Git, Hudson and Redmine. The blade server is front ended by a Supermicro server that insulates the blade server’s integral ethernet from external ethernet traffic, and we VNC into the front end to access the bades’ ILOs (to turn the blades on and off as need be) and to SSH into the blades for diverse reasons. Perhaps one day we’ll have a play with Hadoop and Lucene on a second blade server.

Hardware RAID arrays: We know and like Adaptec 3000 and 5000 series cards which we typically use configured to provide RAID 6 arrays, which provide for two simultaneous disc failures in an array. For these RAID arrays we use enterprise h SATA drives, namely 500GB Western Digital RE3 drives. These provide certain RAID-related advantages over other SATA drives.

SSDs: Only one way to go at the moment in the SSD market: OCZ Vertex 2 SSDs do something like 50,000 IO operations/second.

Monitors: Gotta be IPS or PVA film LCDs for superior viewing experiences, none of this cheap TN matrix stuff we spend hours each day looking at displays.

That’s it folks!

Personalisation is the supply of services and/or data based on a model of a user.

Supply and model are interpreted broadly, to mean by a machine or human agent. Examples of human agency are personalised shopping, by one person for another,  and self-regulated learning, by someone for him/herself.

Not only do we have to deal with future fiscal constraints, but we also have to deal with a future in a world that is characterised by ever-increasing change and consequent levels of uncertainty. We need the ability to both learn in changing environments, and to apply the results of that learning in those changing environments.

UK learning practice suffers from systemic problems in the school education system that are centered around teachers being forced to produce students who can pass tests and exams. While doing well is not wrong, the by product of this kind of education is that children loose their natural curiosity and instead become set in a pattern of doing the minimum that they think is needed to achieve a desired grade, most often by simply reactively doing what they are told to do by a teacher. By the time learners enter HE they are fully habituated to this pattern, and most often HE only further institutionalises the problem.

Instead, we should be interested in the construction of a personally relevant learning experience for each and every learner, such that learners gain characteristics that are useful in, and lead to, social, cultural and economic development.

And while we value the existing utility provided by learning support systems, we note that they fall far short of the mark in supporting this kind of education particularly on the level of the national massification that we need. In fact, learning support systems are themselves incapable (for the forseeable future, at least) incapable of supplying this kind of education.

Could teachers and lecturers instead supply these personally relevant and meaningful learning experiences? Diana Laurillard, in her inaugural professorial lecture at the Institute of Education, points to the impossibility of teachers on their own supplying a personalised learning experience.[1] Indeed, the impossibility of supplying learners with personalised learning experiences is, in general, a self-evident truth to those of us who work in the education sector.

The only course that remains, then, is for learners to construct these learning experiences for themselves. In fact, self-directed learning is the only economically feasible means of providing a personalised and meaningful learning experience on any kind of massified scale.

What does it mean to be a self-directed learner? Essentially a self-directed (or ‘self-regulated’) learner takes control of his or her own learning, and in the process performs various metacognitive activities that involve thinking about their learning. For example, a learner might set learning goals, and then perform, monitor, replan, etc their learning activities, possibly changing their overall or intermediate learning goals in the light of ongoing learning performance and newly gained knowledge. It is hard to divorce this view of learning from learning together with other learners: Learners will learn best when applying self-directed learning as part of a group of learners who are assisting each other in learning together.

While this sounds radical in terms of the staus quo, it is important to note that the central aspects of institutional education – syllabi, assessment and formally appointed teachers – are all eminently possible within the approach. We do however recommend that the nature of assessment changes to formatively support the processes of learning and learning to be a self-directed learner.

Where is the support for socio-economic change mentioned in the first paragraph?

“Self-management of thinking, effort, and affect promotes flexible approaches to problem-solving that are adaptive, persistent, self-controlled, strategic, and goal-oriented.” [2]

These are precisely the skills that we need to deal with a changing world: Experience gained in learning within changing knowledge landscapes will create persistent transferrable skills that underpin life long learning, and life long learning skills are essential to enact the kind of change we need.

So specifically there is a proposal here, that we have to transform the learning practices of learners in higher education. This is a vast pedagogic shift that is bound to hit innumerable barriers, but one that, I posit, is inevitable if we are to successfully deal with real-world constraints on education and at the same time transform our longer-term socio-economic prospects.

So how do Personal Learning Environments (PLEs) fit into this? Whatever the brouhaha about what a PLE is, the thing that underpins and unifies the PLE movement is that it is about learners doing it for themselves; learners taking control of, directing and managing their own learning. A PLE provides the infrastructure for that kind of learning. Of course, infrastructure is only part of the solution; the other part is achieving the pedagogic revolution. This relies on students unlearning their current learning practices and adopting new practices in a guided and supported fashion.  Sometimes I privately refer to this as “achieving the impossible” when considered across the HE sector, or even within a unit as small as a three-year degree programme. There are profound impediments to achieving this that are rooted in institutional culture and practice.

But despite the barriers and my sometimes thoughts “impossibility”  I remain an optimist; there seem to be two approaches that may pay dividends. Firstly a grass roots  approach of individual university  departments committedly adopting an approach to self-directed learning and working towards this goal over several years, concentrating initially on key steps to change student learning culture. Secondly, there is a more formal funded cross-institutional approach where one could imagine a programme that plans and delivers roll-outs of materials, methods and pedagogy produced by early adopters, feeding back into the programme experience from both adopters and adoption/roll-out processes.  Neither approach provides an easy path, but given that pressures on education seem to increase unabated, the right drivers for these approaches may have arrived. And while these are likely  long-term activities where there would likely be no short-term effect on the current economic climate, I fully believe that such a programme would have profound positive implications for our national future.

[2] Paris, G. and Winograd, T. The Role of Self-Regulated Learning in Contextual Teaching: Principles and Practices for Teacher Preparation, http://www.ciera.org/library/archive/2001-04/0104parwin.htm

The project is aimed at three targets:

1. Putting a large portion of Fishbase’s relational data on the web as Linked Data triples. We anticipate, during the next nine months, putting more than 100M triples on the web. This has its own challenges, eg converting the relations we are generating (using R2D) with existing standards, making sure that copyright information is maintained, and not least of all, dealing with a large data-set.
2. Building basic species pages from linked data, and providing a self publishing mechanism for collection and aquarium data
3. Adding a live tutorial medium for SPARQL based on our open source iDoc documentation product

This project has been running for about a month, and we are making good progress with data conversion and have already constructed basic species tables from data in a triple store (we are using 4store with a custom Ruby on Rails front end for this).

The fishBase team consists of Bijan Parsia (UofM SCSc), Sean Bechhofer (UofM SCSc), Rainer Froese (IFM-GEOMAR and FIN Inc), Dave Workman (Hedtek) and Mark van Harmelen (Hedtek and UofM SCSc).

The fishDelish blog and project site can be found at http://fishdelish.cs.man.ac.uk

The project will  investigate student authorship of textbook material: “OurWikiBooks will undertake co-development, with teachers and GCSE and A-level students, of a new digital collection of key concerns and knowledge in computing education.”  As part of this we will address teacher continuing professional development and the development of their identity as computing teachers.

iDoc is an open source community documentation system for the production and use of help documentation for web sites and applications. It’s more than just a vanilla documentation system:  We believe is important is for the users of an application to be able to contribute back and help others learn to use an application application — iDoc allows this by allowing users to sign up and comment on help pages. It is also possible to set iDoc to allow any user to create or edit help pages, indeed, this is currently the default behaviour!

Why another documentation system?

There were several motivations that inspired this project.

The foremost reason was that we wanted an easy-to-use documentation system for the Manchester PLE, where our desire was to host very short (from a few seconds to something less than a minute)  video’s to help newcomers use the system. This isn’t to imply that the PLE is hard to use, but rather that some potential users will have no idea, e.g., what a blog is, nor how to use blogs within the system. Equally, we don’t see ourselves producing all the documentation, rather we expect some to be produced and/or improved by our communities of users.

Most documentation systems we investigated involved the use of some specialised markup to produce the content from some other source. This isn’t desirable for a system geared towards general users of an application, as they may not be technically skilled, and, even if they were, we imagined that they most likely wouldn’t want to learn yet another syntax for creating content. It’s for this reason that iDoc provides WYSIWYG editors to create pages and add comments. You can even use the WYSIWYG editor to embed videos from YouTube and Vimeo in both help text and comments, and this will be expanded in the future to provide for videos from other sources.

The second reason was to test a new development approach for use within Hedtek. The approach we are now using is user story-driven approach, Behavior Driven Development (BDD). In short there are various advantages to BDD, but pretty much its a scenario-based design technique, which leverages the automated system development testing espoused by Test Driven Design (TDD) and enables accurate estimations of the development resource needed to implement the system under consideration. Also importantly, BDD uses iterations in the development, in a similar way that SCRUM uses sprints to deliver working systems at the end of each sprint. We wanted a smaller (but non-trivial and useful) project to get up to speed and make sure we were comfortable in the process before using it on larger projects where a new process could simply be one more unknown that might add problems to the development process. We are glad to say that BDD worked tremendously well for us and is now our default approach for all new projects we undertake.

Finally, there was the desire to have a non-trivial project developed in a carefully engineered way as an example in a new Software Engineering course we are developing for and delivering in the Advanced Master’s programme in the University of Manchester’s School of Computer Science.

The current iDoc status

After two of four iterations, iDoc has reached the stage where it provides useful documentation facilities. It provides a clean URL syntax for pages, WYSIWYG editors for content, and drag-and-drop editors for positioning pages in a documentation tree. Users can sign up, develop and publish page content, and comment on pages. There are some administration features that can, if so desired, limit who can create and edit help page body content, and there is the ability to select community moderators who can remove unsanitary comments from comment lists. There is also search functionality to more easily find relevant parts of the documentation, backed by Xapian search so the search is capable of case-insensitive searches, wildcards, boolean queries and all the other niceties of an advanced full-text search.

Technologies

iDoc has been developed with the latest stable Ruby on Rails stack (version 2.3.5). It requires backing by a database, and was developed using a PostgreSQL database. However, iDoc is currently relational database agnostic. The site uses Haml to provide HTML templates, and Compass to provide semantic CSS. Development made use of several Ruby and Rails development tools, including RSpec, Cucumber and Webrat. For dependency resolution, iDoc was recently migrated to make use of Bundler, a new Ruby system for specifying and installing libraries using the RubyGem package manager. It should be easy to install iDoc onto a system with Ruby and Xapian installed, and deployment can be automated with Capistrano. Current installations of iDoc are served up with Apache and Passenger, although again there shouldn’t be any problems with other Rails server technologies, such as Mongrel, Thin, Unicorn or even a plain old FastCGI server.

For Xapian integration, a new Ruby gem, Xapor, was developed to allow for the specification of Xapian indexes very easily through ActiveRecord. The interface is inspired very much by other popular search plugins, such as ThinkingSphinx. Xapor is currently at version 0.1.2, with the features required for iDoc currently implemented. It is intended to continue development of this gem to provide a much more sophisticated search integration with Xapian. Xapor currently also relies on the gem XapianFu, which provides an easier interface for Xapian than the plain Ruby bindings but isn’t particularly sophisticated for ActiveRecord integration. I open sourced Xapor on Github and the gem is available on Rubygems.

Future development

There are several more features that we want to integrate into future versions of iDoc. Some of these include:

• Notifications. It would be highly desirable to have notifications available to users. These would include things such as page additions, deletions, updates and new comments. Currently, it is planned that these would be served as an RSS feed, although other notification systems could also be added (e.g. twitter and/or email)
• Published APIs. iDoc isn’t intended as a standalone system, and although it is easy currently to just embed help links or navigate the documentation by hand, it would be much nicer if some degree of automation could be performed.
• An improved user login experience. Login currently relies on registration, but could be extended to include the use of other signup systems, such as Facebook Connect, Twitter OAuth and OpenID, to allow users to re-use their existing accounts on these servers rather than creating a new account.
• Automatic generation of content sections. This is an important improvement, and would improve the navigation of documentation pages. The general principle would be to automatically provide in-page tables of contents at the top of documentation pages, and provide the ability to edit subsections rather than the entire page as a block. [Stop press: Now partially implemented and soon to be rolled out]
• Interface customisation. While this is certainly possible currently, it requires a customiser to hand edit CSS in order to do so. It would be much nicer if there was a web-interface method to customise at least some of the interface elements for an installation, and possibly even on a per-user level at some point in the future.

As iDoc is open source, we hope that others will extend the system in useful ways.

Our last post was way back, during ALT-C. Since then we’ve been totally heads down, working on projects for clients and on our own PLE Project.

The JISC-funded MOSAIC Project, with SERO and co-partners Ken Chad (Ken Chad Consulting) and Paul Miller (Cloud of Data) investigated, at many levels, a new social and personal search approach to library catalogues for learners. This breaks new ground, by providing social search facilities based on attention data that is encapsulated in library loan information. The project and the prototype we provided produced some really good outcomes, with JISC promoting the outputs in various meetings, and commissioning a bit of further work to see how the MOSAIC search approach can be progressed in UK Higher Education. You can try our prototype at mosaic.hedtek.com (this will move to http://demo.mosaic.hedtek.com and I’ll edit this post and remove this comment then).

The iCue project also produced interesting and useful results for client Mimas. We were tasked with the development of an algorithm to help structure the 70M record Copac database, which contains details of the the holdings of some +/-40 UK university libraries. The end aim was to provide a better search experience for Copac users. The standard way of doing this is to use a scheme based on the Functional Requirements for Bibliographic Records (FRBR) approach. Using a three level FRBR approach (such as e.g. used by WorldCat Local)  we developed a FRBRisation algorithm, prototyped and refined it, and provided a first level mockup of structured search facilities.

We’ve also been working hard at improving the Manchester Personal Learning Environment (the Manchester PLE). Last September I used it to support my Interactive System Design course (with a cohort of 80 students attending the University of Manchester ‘s MSc in Advanced Computer Science). This threw up some bugs and identified a lack of support for teachers when the PLE is used in an institutional setting. We fixed the bugs, performed some code refactoring, vastly improved the user interface, and added support for teachers by providing any PLE suer with the ability to set a blog based assignment. Friends Sefol re-implemented the spaces for us, concentrating on reliability. Finally we feel we are ready for a public release.

Dave also implemented the marvelous iDoc system for the Manchester PLE, where we are going to mostly use it to host short (much less than a minute) help videos. iDoc is designed for user generation of help information, and for user comments on that help information. It can also be restricted to ‘expert’ named users writing the main body of the documentation, with others adding the comments only. It can be used with any application, particularly where the authors of that application want help buttons to deep link to particular help information. If you don’t know what deep linking is, the effect is to click a help button for xyz while trying to do xyz, and being taken directly to the help for xyz. But once you’re there, you can browse around the other help information. We’ve decided to open source iDoc. This text will become a link to a post to iDoc.

iDoc was our first venture into Behavior Driven Design (BDD). And, from an engineering point of view, the experience was massively successful. We’ll post on that later. Dave open sourced a ruby gem in the process. We did two of four planned iterations, so there are still additions to come to iDoc. However its eminently usable now, and we have, bar the the copyright notices, an Open Source product in iDoc. Try it out at  demo.idoc.hedtek.com, and get the source via source.idoc.hedtek.com. (Both coming soon, I’ll remove this bracketed comment when these links go live.).

.

Advance notice of The PLE Conference, Barcelona 8-9 June July 2010.

From an e-mail from Graham Attwell, with a mild edit:

The PLE Conference is intended to produce a space for researchers and practitioners to exchange ideas, experience and research around the development and implementation of PLEs; including the design of environments, and sociological and educational issues and their effectiveness and desirability as (informal) learning spaces.

Whilst the conference will include a traditional research paper strand, we also wish to encourage proposals for sessions in different formats including workshops, posters, debates, cafe sessions, hands on sessions and demonstrations. We will also provide opportunities for unconferencing events, including the provision of spaces for informal meetings and discussions.

The face-to-face conference will take place in the Open Innovation Centre for the Community in Barcelona. The Centre located in a converted factory, and provides a variety of purpose designed spaces for exchanging ideas, including an auditorium, meeting rooms and spaces for informal discussions.

The centre is well equipped with multi-media equipment, and there will be free access to a fast wireless network throughout the venue.

The conference will utilise multiple on-line spaces including a web site, a social networking space for participants, and other social software.

Conference themes include (but are not limited to):

• Theories and frameworks for Personal Learning Environments
• Technologies and software for developing Personal Learning Environments
• PLEs in Practice (case studies, approaches to using PLEs)
• Educational institutions, change and PLEs
• Pedagogical approaches to managing personal learning
• The development and management of Personal Learning Networks
• Mobile PLEs and augmented reality
• Supporting informal and and contextual learning
• Using PLEs in organisations
• Using PLEs for work-based learning

The conference organisers will shortly be launching the conference web site and issuing a call for contributions.

For the last year we’ve been a bit inward facing with several large projects to attend closely to, including getting the PLE to a usable state, and running the first trial with 80 users. This blog has been pretty much focused on company stuff, on presentations I’ve made, and on the Manchester PLE.

Wanting something a bit more lightweight, with a bit more of a personal touch, I’ve established a tumblelog of over at hedtek.tumblr.com. Who knows how it will evolve, but right now it seems just right for stuff we like, with a dash of tek sauce.

 
We are coming to the end of the JISC-funded Mosaic Project. As part of that project Hedtek has designed and, with help from our friends at Sefol U.G., implemented a prototype search engine that allows personalised searches of bibliographic data according to one’s own course cohort, or according to various other faceted criteria that allow the adoption of various points-of-view.

Various resources pertaining to Hedtek’s Mosaic search engine appear here:

• Slides from the Mosaic meeting on 18 November 2009. (Other slides from the meeting)
• A brief description of the search engine and its use
• The search engine URL (at the end of this post)

Slides

A brief description of the Mosaic search engine