Semantic Library

As someone who no longer works in a library (but for libraries), I feel that I’ve had more space than ever in the last year to think about the future of libraries, removed from the daily grind of library work. Irony, much? This month, there’s been an absolute flurry of content by Australian library bloggers who are completing a 30 blog posts in 30 days challenge. Not something I could contemplate with much of my June to be offline, but they’ve inspired me to post some of my thoughts after reading their posts.

Advocacy, not acquisitions?

One of the topics that’s come up quite a bit, and comes up whenever I find myself in a group of librarians, is what is the future of libraries? I wrote a recent post about some ideas for public libraries, but what, more generally, are the things that will keep libraries of all types thriving in the future? I’ve seen Aaron Schmidt talk about libraries not focusing on circulating content so much any more a couple of times, and while I think that there is always going to be an acquisition arm to what we do, I wonder if more of us will join in the work to reform copyright, advocate on behalf of users, work on projects to increase access. For many, that would be a shift in skillset and expectations about “what librarians do”. For others, this is what they have been doing their whole career – every time they refuse to sign a an unfriendly publishing agreement, or advise researchers to retain copyright to their work.

Find everything, through collaboration

We need to collaborate more – we have consortia at many levels, conferences where we get together, associations where we work together. But all too often, we don’t know enough about what’s happening at the library down the street. I can attest that I have often been guilty of this – sometimes I would visit other libraries in my city to see what they were doing, and attend their events, but it was all to easy for a year or more to go by without talking to or working with colleagues from the other nearest academic library. Libraries are fragmented – if you’ve ever tried to search all the public library catalogues in your city to find a service or book that you want, it’s not an easy task. No wonder users find ‘The Google’ or ‘The Facebook’ easier. It’s just one place.

I loved the Libraries Australia service when I lived in Sydney – the function I used it for has now been replaced by Trove. I cannot tell you how often I used or recommended this service to users. This type of service is being rolled out in other countries now, not modelled on Trove, but the idea of a single-search portal for all libraries in a country. For example, here’s the Lativan Library Portal.  This is what we should be doing.

Teaching data

There’s definitely a role for libraries in teaching users how to make the most of data. Libraries teach classes for users on databases, and finding articles, using social media, information literacy and so on. But I think there is a strong need now to begin to teach how to work with open data sets, linked open data, to know how to adapt and build on this data, and to contribute data back. For quite some time now, I’ve been thinking about the role of linked open data in libraries, and I want to expand on this in future posts, but the opportunity is two-fold: one, to make use of it ourselves in what we produce, and secondly, to teach others. More and more governments are providing data. The World Bank recently opened up its data set. Being able to use data, present it, and do something with it is becoming more critical, but there’s not really anyone out there filling this need. Why not libraries?

3 ideas in 30 days

I may not have been able to take part in the 30 posts in 30 days challenge, but I thank everyone who is for providing such thought-provoking posts and inspiring me to put a few ideas out there.

Internet Librarian International 2009 is being held in London next week. I will have a presentation at the conference, but unfortunately due to a work travel commitment I am not able to attend myself. Instead, this gave me an opportunity to find a co-author/presenter and explore some  ideas about the role of the catalogue in discovery. I’ve called it, “Adaptability, aboutness, and authenticity: Towards discovery platforms with next-generation catalogues and Linked Data” -

>Next-generation catalogues give libraries an opportunity to better meet user needs and changing search habits, but there is a need for more than innovation at the interface level. Many academic libraries have now implemented software that provides an improved search experience in the catalogue, but libraries also need to be able to anticipate user needs and emerging uses of resources and to provide tools that meet needs one step ahead. Linked Data, associated with the Semantic Web, is one way to open up data to the web to expand the ways that users can be connected to library resources. Linked Data also provides the opportunity to consider how catalogues present resources to users. The paper presents a case study of Work Based Learning as an example of a multidisciplinary academic field in which the notion of authenticity is evolving. The question of whether library catalogues can or should help navigate users to authentic resources is considered.

We enjoyed writing it and thinking about how subject experts and librarians may be able to work together to create new platforms for sharing data within and beyond libraries.

I’ll be following ILI2009 tweets from afar, if you’re going have a fantastic time and learn well!

Last night I was watching a few news clips on the BBC News website, and at the end of each video was links to “more like this”. I made a comment on Twitter:

“End of BBC news videos on their site suggests ‘more like this’. Don’t really want to watch ‘more like this’ if I’m watching sad news.”

Jodi Schneider responded:

“@blisspix That suggests mood-based recommendation for news outlets. Interesting! “More like this” “Opposite mood” etc.”

Now that’s an interesting idea, and it got me thinking about recommendation systems in general. Mostly I’ve been unsatisfied with everything from Amazon to iTunes Genius. Their algorithms are, to my mind, too blunt.

Etsy’s explore options (colours, pounce, time machine, geolocator, etc) are an amazing leap forward into more meaningful browsing and recommendation. Whether you are into handcrafts or not, this site is an excellent example of making the most of metadata.

But we can go much further, radical recommendations, and I think it’s something librarians can and should be involved with. We have a longstanding strength in readers’ advisory and I think you’d be hard-pressed to find a librarian who has not heard of Nancy Pearl’s Book Lust.

There are other efforts already to provide online recommendations on what to read next, like ReadMe from MetaFilter, What do I Read Next? from Gale, and both human and automated recommendations in LibraryThing. Yet they are not at the depth I would like to see – the depth that people like Nancy Pearl provide. Now to be clear, I don’t want to replace humans and the strength of emotion and individuality in recommendation, but I do want a more reliable way to explore what I should read, watch, or listen to next. And that’s likely to be people-generated. I do want a recommendation that will take into account mood, but also things such as -

• Am I buying this item for the library? Then I care about how durable the book is, how up to date it is, how relevant it is to courses.
• Am I buying this item for me, to read or listen to on a train/plane trip/while sitting by a pool on vacation? I’m going to want something different for all of those.
• What are the instruments being played in the music? I don’t really like auto-tune or drum machines, for example.

The very human nature of recommendations though, is exactly what makes it so difficult.  I find music recommendations particularly troublesome, whether from AllMusic, Amazon, iTunes Genius or Last.fm. They are usually too broadly applied to really be of use or don’t reflect how I feel. For example, the ‘mood’ for my favourite band, Radiohead, includes the following:

• Cold
• Epic
• Sprawling
• Austere
• Paranoid
• Tense/Anxious
• Cathartic
• Distraught
• Insular

And I don’t feel that way at all when I listen to their music, so these options really don’t help me.

How can we use the strength of individual listeners and readers? And how can this be done while protecting privacy? In libraries, we have been stepping into recommendations at the catalogue level only tentatively, both because we don’t actually have any of that data most of the time and to protect the privacy of individuals. Is there a way to collect deep, meaningful recommendations about people who like particular books + videos + websites + information without creating a dangerous profile? That will be a key challenge of the Semantic Web.

So how can recommendations be improved, made more granular, more relevant? Small, distributed pockets may be a way. Information collected at the listening level (eg last.fm, iTunes Genius) could be queried against opinions of dozens of reviewers, and hundreds of fans, and against statistical data like “bands that have supported this band on tour”, “bands that covered this track” (available in MusicBrainz ) and information about what guitars they play with. There shouldn’t be just one entry on AllMusic for a band, but rather many different pieces of information scattered across the web, connected using Linked Data giving strength and flexibility to be able to say “I am looking for happy music made in a cold climate in the mid 90s with slightly political lyrics that has a similar instrument profile to Ride” and in the next minute want to listen to dance music, and be able to find just what you’re looking for.

Much of this is part of the vision for the Social Semantic Web, Alexandre Passant from DERI has a great slidedeck on this topic:

2 Comments »