Linked Open Data – Melbourne Workshop

Mia Ridge being in town provided the initial impetus to organise a Melbourne LODLAM event. On April 17th apx 35 people from a range of sectors, including memory organisations, tertiary institutions and government departments gathered at the Melbourne Museum. It was a lively session and concluded with agreement to continue discussions focused on 2 LODLAM projects: place names and World War 1.

The following write-up is from notes taken by Ely Wallis (@elyw) and myself Eleanor Whitworth (@elewhitworth).

LIGHTNING TALKS

With a focus on practical and pragmatic applications and opportunities for sectors to work together we started with a series of lightning talks.

Melbourne LODLAM speakersMia (@mia_out) opened with an overview of LOD & LODLAM, including ‘quick wins’ and common issues. We noted in particular:

  • The importance of being clear with others in your organisation about what it is you are making available – the metadata or the content – to alleviate concerns.
  • Just making your metadata freely available is a good start, especially if copyright issues prevent you from making images available.
  • Re licensing, make the data available for reuse with no licensing requirements so there’s no need to go back and negotiate a license with the source organisation. And, in terms of openness vs usefulness, allow commercial re-use to enable external developers to meet user’s needs (in ways your organisation may not be resourced to do).

Discussion following Mia’s talk included a question about releasing data and crowdsourcing corrections for it. Mia noted the need to consider how the corrected data will be ingested back into the collection management system. In response to a second question on how APIs square with LOD, Mia commented that the downside to APIs is that it’s difficult to include all the links to authority files and vocabularies. So, they’re good for sharing data between local data stores, but often semantic linking is lost in the transfer.

Mia’s talk notes can be found here, and a LODLAM reading list here.

Peter Neish, Systems Officer, Victorian Parliamentary Library, (@peterneish) talked about the use of OpenCalais to automatically tag media releases harvested from websites. Peter noted that as OpenCalais is a Thompson Rueters product and uses its own (media company) ontology, it has a good vocabulary for media personalities but not so good for historical people, or Australian places or events. Peter noted the tagging has been relatively good with 85% of them being usable and useful. There was a suggestion in the discussion that followed that an Australian version of OpenCalais might be in development by Fairfax Media and Macquarie University, but we couldn’t find any references to it…  Peter’s slides are up on Slideshare: Semantic Web Technologies at the Victorian Parliamentary Library.

Helen Morgan, Research Fellow, eScholarship Research Centre, The University of Melbourne, talked about “Linked Dead Data” and how well-formed URLs in websites, authority files etc can be useful research tools. The e-Research centre has developed the Online Heritage Resource Manager which includes a variety of public knowledge resources, including the Encyclopedia of Australian Sciences, the Australian Women’s Register & the Australian Trade Union Archive. Whilst these online archives return some dead links giving 404 errors, the Wayback Machine and Pandora are useful resources for researchers to track back. Helen noted that the whole link is the core of citation, and that shortened links, whilst they serve a purpose (e.g. on Twitter), are hard to keep discoverable as there’s nothing in the shortened link that gives a clue to what the long URL might have been.

In the discussion that followed, Forbes Hawkins @forbeshawkins suggested that we think about styling dead links up front (eg. flag that the link is dead and direct the user to the Wayback machine or Pandora) to assist users. It was also noted in the discussion that it is important to keep the context of data in order to not pre-empt what the user may want from the data.

Conal Tuohy, eResearch Business Analyst, Victorian eResearch Strategic Initiative (‏ @conal_tuohy) presented on the Humanities Networked Infrastructure (HuNI, pronounced honey) project. Following some ‘sweet’ jokes, Conal outlined how the HuNI consortium is setting up virtual laboratories and making a range of data sets available from providers at different institutions, with different subject domains, hosted on different software platforms with various levels of LOD-ness. As a lot of HuNI providers have their own vocabularies, a large part of the work will be mapping. Conal’s slides are up on Slideshare: HuNI

Adam Bell, Manager Web Production, Australian War Memorial (@bumphead) outlined how the AWM is planning to set up a rich semantic tool with a focus on the WW1 centenary and is working with web scientists (eg at the CSIRO ) as well as the cultural sector to work through the issues. They are also versioning ontologies and using the building of ontologies to get librarians and archivists involved. Adam noted that Drupal has good taxonomy tools and that they use it as a proxy to other services. The AWM is also looking at ways to make their website more social. Adam noted that whilst these social aspects (eg. richer faceting and better ways to encourage people to explore the site and find information about their relatives) are not strictly LOD, LOD helps to achieve these functionalities for users.

A SURPRISE VISIT

A spontaneous international drop-in by Jon Voss and Simon Sherrin via Skype provided the perfect lead-in to the general discussion…

Jon Voss and Simon Sherrin

Jon Voss and Simon Sherrin

Jon put forward the following:

  • You need to have open data before you have linked open data.
  • Make your metadata open if you can’t make your content open.
  • Link where you can.
  • There are international LOD projects around WW1 and the Olympics underway.

GENERAL DISCUSSION

  • Ely summarised the commonly expressed issues in the session as being: dead links, sensitivity re clean/messy data and managing ontologies and vocabularies.
  • Mia noted that it’s great to have a place to go to ask questions and find out how others solved these issues, and that Europeana has provided this critical mass, but that a local group may be useful.
  • There was a general discussion around the treatment of Indigenous material (eg. the importance of systems to control access). Mia noted that Australia is in a good position to lead other regions such as Europe in this area.

WHERE TO FROM HERE

It was agreed to continue a Victorian meet-up with a focus on 2 projects:

1)  Place names: State Library of Victoria has digitised a range of Victorian parish maps. These could be mined for place names. The AWM is using a geonames service to build place names into an ontology and could share this.

2) WW1: Share information and look at opportunities to link WW1 data.

Eleanor will facilitate a follow-up meeting. If you’d like to be involved contact her at: eleanor.whitworth@dpc.vic.gov.au or @elewhitworth

IN CONCLUSION

A Twitter discussion between Claudine Chionh @claudinec and Mia nicely summed up the morning: #lodlam links people too!

Melbourne LODLAM muffin remnants

Melbourne LODLAM muffin remnants

This entry was posted in Uncategorized and tagged , . Bookmark the permalink.

2 Responses to Linked Open Data – Melbourne Workshop

  1. Pingback: Keynote: ‘From Strings to Things’, LODLAM Melbourne workshop | mia ridge

  2. Pingback: Upcoming talks and travel | mia ridge