paul walk’s weblog

I was pleased to be invited by Brian Fuchs to a ‘Million Books Workshop’ at Imperial College, London last Friday. A fascinating day, in the company of what was, for me, an unusual group of 20-30 linguists, classical scholars and computer scientists. The morning session consisted of three presentations (following an introduction from Gregory Crane which I missed thanks to the increasingly awful transport system between London and the South West) which brought us up to speed with some advances in OCR, computer aided text analysis and translation, and classification. The presentations were intended to form an ordered progression:

1. From Image to Text: OCR and Mass Digitisation (Dr. Thomas Breuel, DFKI and Technical University Kaiserslautern)
2. From Text to Information: Machine Translation and Syntax Recognition (David Smith, Johns Hopkins University, & David Bamman, Perseus Project)
3. From Information to Learning: Machine Learning and Classification Techniques (David Mimno, U Mass, Amherst)

Listening to these presentations, I quickly found myself well outside of my comfort zone, in terms of both the science and the domain (classical literature), so it was a challenging and exhilarating morning! It was difficult to take comprehensive notes as I had to really concentrate on the presentations at times in order to follow them - the pace was pretty smart, with jargon and ‘in jokes’ galore.

David Smith, Johns Hopkins University gave a fascinating and entertaining presentation which outlined some of the challenges, and advances, in language parsing and translation. He pointed out that although the structured view of the semantic web is a seductive one, even the newer online, digital genres such as email, blogs mostly use unstructured or semi-structured text. However, parsing free text is very difficult, especially with the growing scale and diversity of texts available on the web. To illustrate this he employed a series of (sometimes amusing) translations from the Google translation service. The best available technology today uses supervised machine learning techniques to build a treebank. An alternative approach employs semi-supervised, modelling techniques. Parallel texts in different languages are useful but, for some languages, only the Universal Declaration of Human Rights exists as a parallel text! As an aside, David pointed to the potential advantage in search engines searching several languages: if you enter your query in English for example, by searching resources in other languages, the search engine automatically expands the search, utilises synonyms etc. ‘for free’. This can then be more effective than monolingual searching. David offered a future based in pragmatism: translation support rather than fluent translation.

David Mimno presented on classification, sequences & topic modelling. In an interesting talk, it was the visualisation (as a topical transfer graph) of topic relations extrapolated from citations in a set of scholarly communications which really got the audience engaged - a series of questions ensued before David could move on. He illustrated his work with accessible examples: for example, it turned out from one experiment that the single term most likely to identify email spam was, believe it or not, the word “red” showing in the markup, owing to the fact that “only spammers use red text”! Apparently, he had a system which could classify any of Shakespeare’s plays as tragedy, comedy or history…. with the exception of Romeo and Juliet, which comes out as a comedy for some reason….

The takeaway for me was that some of the technology in these spaces is maturing. Thomas Breuel, for instance, made a compelling case for really effective OCR (Optical Character Recognition) in his description of the open-source OCRopus project, which he leads and which is sponsored by Google. Building on previous systems like the character recognition system Tesseract, OCRopus employs a modular design with components which offer the following workflow, focussing on the processing of scanned books:

layout analysis -> isolated character recognition -> statistical language modelling - > text

The project is heading towards a beta release this year, and the team plan to create a deployment ‘bundle’ in the form of an Amazon EC2 AMI. I didn’t quite catch the details but I think they have found a way to monetise this through the Amazon referral program, which sounds interesting. In any case, the idea is that one could take the AMI, deploy it, run it for a few hours to process a particular scan, and then shut it down again - potentially a very cost-effective way of proceeding. Thomas made the point that, as OCR technology continues to improve, we are likely to want to process scans of books several times. He explained how the project was aiming towards a “full digital library pipeline”, a system which could be deployed from a connected laptop: with the new affordability of powerful digital cameras, a researcher might photograph a book’s pages themselves before feeding the resulting image into the OCRopus workflow OCRopus can handle the distortion effect of non-flat pages very effectively). Another interesting aspect of this work is the distributed parallel training which underpins the statistical language modelling: a large model is achieved by combining many little models created by many people, through the web. If you are interested in this area, then you should also check out the hOCR format specification and tools.

I had been invited to this workshop because of my role and interest in the deployment of services at a community and network level. I joined a panel at the the very end of the day where we were invited to consider what services and infrastructure might be required, in the UK, to support the digitisation and useful processing of a ‘million books’. We didn’t get very far with this because we had run out of time and, I suspect, energy by this point, but the question remains…. I’ll be picking this up with some colleagues in due course.

Fascinating day, and topped of with a quick pint standing outside a packed London pub in a light drizzle, which was actually a refreshing and pleasant way to conclude!

Tags: David A Smith, David Mimno, Gregory Crane, hOCR, millon books, OCRopus, Thomas Breuel

In Repositories thru the looking glass over on the eFoundations blog, Andy Powell gives a summary of a keynote he gave to the Vala Conference last week. It’s interesting stuff, and I will take the time to look at the presentation slides as well. I mostly agree (vehemently in some instances) with Andy’s points, though I do find myself questioning some parts of this, so I’ll quote some snippets and make a few comments here.

Firstly, that our current preoccupation with the building and filling of ‘repositories’ (particularly ‘institutional repositories’) rather than the act of surfacing scholarly material on the Web means that we are focusing on the means rather than the end (open access)

It’s hard to deny that there is a current preoccupation with establishing repository systems of one kind or another and populating them with content, and also that there is a focus on institutional deployments. However, I’m not convinced that open access is (or at least is going to remain) the sole driver behind the development of institutional repositories. From an institutional perspective, it absolutely makes sense to want to manage the outputs of research conducted within the auspices of that institution.

A common use for an institutional repository is to house eprints. Were it not for the open-access imperative, we might have expected software designed to manage eprints to fall somewhere between a document-management and a content-management system - both familiar to a large number of institutions. I think it is interesting that it might be considered to be open-access which has skewed the development of repository software in some respects - the community has largely started from scratch, building repository software, where it might have made more sense to simply adapt what was there.

So I half agree with Andy - we do seem to be focussed on the means, but I think I am sympathetic to those (institutions at least) who find themselves pre-occupied with this.

Secondly, that our focus on the ‘institution’ as the home of repository services is not aligned with the social networks used by scholars, meaning that we will find it very difficult to build tools that are compelling to those people we want to use them. As a result, we resort to mandates and other forms of coercion in recognition that we have not, so far, built services that people actually want to use. We have promoted the needs of institutions over the needs of individuals.
Instead, we need to focus on building and/or using global scholarly social networks based on global repository services.

There are four sentences here, and I completely agree with the first three and a half! I find myself wondering who ‘we’ are in this. Now that institutional repositories are becoming a reality, the ‘we’ is going to expand to include people who simply have institutional interests - who have no real interest in open-access for example beyond it being a requirement for them to support. The MIS Manager of your average institution, for example, will start to get involved once institutional repositories get embedded into the business which is a university. The half sentence I don’t quite buy is the “global repository services”. Why can’t we “focus on building and/or using global scholarly social networks” (which I support) based on institutional repository services? We don’t have a problem with institutional web sites do we? Or institutional library OPACs? We have certainly managed to network the latter on a global scale, and built interesting services around this….

Finally, that the ’service oriented’ approaches that we have tended to adopt in standards like the OAI-PMH, SRW/SRU and OpenURL sit uncomfortably with the ‘resource oriented’ approach of the Web architecture and the Semantic Web. We need to recognise the importance of REST as an architectural style and adopt a ‘resource oriented’ approach at the technical level when building services.

Absolutely - couldn’t agree more. Yesterday, at a JISC committee meeting, I argued that a resource-oriented-architecture and the service-oriented-approaches being encouraged by the e-Framework could complement each other if intelligently and judiciously applied. Incidentally, last Friday, I attended an excellent CRIG workshop devoted to exploring the relevance of ReST to repositories. Matt Zumwalt of MediaShelf showed a working ReST interface on Fedora, and Oxford University’s Ben O’Steen used this to develop a client app, in real time, in Python.

I think we agree that the individual’s interests may often be orthogonal to those of the institution. This may have always been the case but it is, perhaps, increasingly an issue as recent developments and trends on the Web empower the individual at an accelerating rate. I wonder if the user-centric/institutional/global debate around repositories is just symptomatic of a tension about to become apparent all over the (institutional) Web?

Having said all this, when visiting the outer limits of repository software development, I am occasionally reminded of the Knight:

‘I see you’re admiring my little box.’ the Knight said in a friendly tone. `It’s my own invention — to keep clothes and sandwiches in. You see I carry it upside-down, so that the rain can’t get in.’
‘But the things can get OUT,’ Alice gently remarked. Do you know the lid’s open?

(from Alice Through the Looking Glass, via Project Gutenberg)

Tags: Andy Powell, Ben O'Steen, Matt Zumwalt, repositories, ReST

Disdain for Google on the part of some academics is not new, but Tara Brabazon in her inaugural lecture at Brighton University, has created something of a stir. Alexandra Frean, Education Editor for The Time Online says:

Google is “white bread for the mind”, and the internet is producing a generation of students who survive on a diet of unreliable information, a professor of media studies will claim this week.

Meanwhile, Andy Powell over at eFoundations counters with:

Blaming the Internet for “a generation of students who survive on a diet of unreliable information” is a bit like blaming paper for the Daily Star.

Indeed.

What I find interesting is the implicit notion of a ‘Google generation’. When I hear this term used pejoratively, I tend to substitute the phrase ‘the trouble with the youth of today….’.

In a wonderful coincidence of timing, I note, via the JISC news feed, that the ‘Google Generation is a myth‘. Apparently, there is some evidence to suggest that age/generation is not an indicator for particular research behaviour:

The report ‘Information Behaviour of the Researcher of the Future’ also shows that research-behaviour traits that are commonly associated with younger users – impatience in search and navigation, and zero tolerance for any delay in satisfying their information needs – are now the norm for all age-groups, from younger pupils and undergraduates through to professors.

Hmmmm.

Caveat: I like white bread - it makes the best toast. Also, I found all of the material I’ve read so far about this via Google (apart from the stuff delivered to my RSS reader).

Tags: eFoundations, Google, Google Generation, Tara Brabazon, white bread