Working with text: Tools, techniques and approaches for text mining

Text mining tools and technologies have a long history in the repository world, where they have been applied successfully for a variety of purposes. These vary from pragmatic aims such as enabling document search and browse facilities, linking related documents, identifying copies or facilitating the deposit process, to support tools for academic research. The latter category includes supporting research on the basis of a large body of documents, facilitating access to and reuse of existing work, and connecting the formal academic world with areas such as the traditional and social media. Research areas as diverse as biology, chemistry, sociology and criminology have seen effective use made of text mining technologies.

However, the uptake and hence the impact of these tools has been uneven. Several obstacles to development and deployment are frequently cited, including the maturity, complexity, and in some instances cost of software packages, as well as scarcity of relevant technical skills. Text mining methods and tools can be fragile and complex, requiring significant set-up time and effort. Projects making use of text mining may also suffer from legal obstacles, such as copyright and intellectual property considerations. The benefit to be gained from deployment of text-mining tools in areas such as institutional repositories or as a research tool in its own right may be difficult to predict without a costly pilot project.

Authors are invited to submit original, unpublished chapters describing research in relevant areas and/or reviewing relevant literature and trends.

Topics of interest include, but are not limited to:

• Discipline-specific research involving text-mining: bioinformatics, chemistry, the social sciences, etc.

• Techniques in text mining: sentiment analysis/subjectivity analysis, opinion mining, affect analysis, metaphor analysis, etc.

• Legal and ethical aspects of text mining/analysis.

• Current developments in text mining.

• Metadata extraction from document text, including formal and informal metadata: ontology extraction, document indexing, document classification, and evaluation of metadata quality.

• Text mining for document categorization or summarization.

• Text mining over the social web: community detection, timelines, etc.

• Evaluation of text mining tools, open-source or commercial: case studies and findings.

• Procurement and evaluation of text mining tools.

Submission

Chapters of 4,500-9,000 words in length should be prepared in either Word or LaTeX. As chapters will be reformatted during the publication process, authors are advised to concentrate on content rather than formatting. Please include any images/graphics as separate files; images/graphics should be 300dpi or better and designed to be readable when printed in greyscale.

Files should be submitted by email to Emma Tonkin <e.tonkin@ukoln.ac.uk>. All submitted chapters will be reviewed on a double-blind review basis.

Important Dates/Deadlines

24-Dec-2012 Title/Abstract submission for preliminary approval

4 -Jan-2013 Author notification

11-Feb-2013 Manuscript submission deadline

24-Feb-2013 Author notification

Publisher

This book is scheduled to be published in 2013 by Chandos, a leading international publisher with specialisms in Library Management, Information Management, Social Media and the Web; it will be distributed in the United States via the American Library Association. It will be available both as a printed publication and as a freely available Open Access resource, increasing the visibility of the final work. For additional information regarding the publisher, please visit http://www.woodheadpublishing.com/en/ChandosInfo.aspx.

Please note: although this book is to be made available as an Open Access resource, authors/contributors will not be expected to pay a fee.

The week begins with a 60-minute talk on “Open Practices for the Connected Researcher” which will take place on Tuesday 23 October at the University of Exeter as part of its series of Open Access Week events.

Brian will give a 30-minute talk entitled  “Open Practices and Social Media for the Connected Researcher“  on Thursday 25 October at the University of Salford as the featured invited presentation for the University’s celebrations for Open Access Week.

Finally Brian Kelly and Ross Mounce, a PhD student and Open Knowledge Foundation Panton Fellow at the University of Bath, will launch the social media programme for researchers with a 60-minute session entitled  “Open Access and Open Practices for Researchers“.

The slides to be used in the presentations will be available with a Creative Commons CC-BY licence. In addition the slides for the presentation at the University of Exeter are available on Slideshare.

A talk entitled  ”What Does The Evidence Tell Us About Institutional Repositories?” has been accepted for presentation at the Internet Librarian International, ILI 2012 conference. This talk, which will take place in session B203 on  Evidence and impact, will build on the ideas described in our paper on “Can LinkedIn and Academia.edu Enhance Access to Open Repositories?” as well as  the paper on “Open Metrics for Open Repositories” by Brian Kelly, Mark Dewey and Stephanie Taylor of the ISC at UKOLN together with Nick Sheppard, Jenny Delasalle, Owen Stephens and Gareth Johnson. The ILI 2012 presentation will also provide an opportunity to present additional findings from our RepUK work, together with our effort with UK Repositorynet+.

Thom explained that the Innovation Zone is a JISC-funded initiative focussing on supporting developments in repository infrastructure in the UK, managed by the Innovation Support Centre at UKOLN and the RepositoryNet (RepNet) at Edina. Support is within four main areas:  technical knowledge-exchange through expert workshops;  sharing of key information on repository components and use cases via a knowledge base;  trials of APIs with developer communities through DevSCI; and the incubation of prospective services, an area currently under development and available soon.

Thom is keen to hear from anyone who has repository-related service with an API they would like to trial with developers. The Innovation Zone is able to offer help in putting you in touch with developers and making links with other projects in complimentary areas of work.

The incubation aspect of the Innovation Zone support will be available soon and can help with new development initiatives such as repository infrastructure innovations, prospective components and microservices. Again, Thom is keen to hear from people who have projects that could benefit from incubation.

To find out more and engage with the  Innovation Zone, leave a comment here and/or contact Thom.

Brian Kelly of UKOLN ISC presented a poster in the Poster Minute Madness session on Tuesday, promoting the paper “Can LinkedIn and Academic.edu Enhance Access to Open Repositories?” which he co-authored with Jenny Delasalle of the University of Warwick. The poster focusses on the importance of adding ‘Google juice’ to your institutional repository by generating more links to individual papers deposited in an IR.

Brian has the largest number of downloads from OPUS, the IR of the University of Bath and he has an h-index of 11 for his papers on accessibility in particular are being well-cited. Research carried out by Brian and Jenny suggests that the large number of downloads and citations may be due to inbound links from popular services such as LinkedIn and Academic.edu.

More research needs to be done in this area, but should repository managers be acting on the current findings? There are obvious benefits of actively encouraging researchers to link to their papers from popular profile services used by their fellow researchers. Jenny’s review of the sector suggested repository managers are not being pro-active in promoting the use of such services. Why is this? What, if any, are the barriers?

Brian wrote a blog post that summarises the paper and another about the poster session. Jenny has blogged further thoughts on the original paper as part of this ongoing discussion.

For those not able to attend or wanting another look, there is a SlideShare presentation available,  based on the poster.

The debate carries on, so if you didn’t have time to contribute during the session or you weren’t able to attend the conference, please join in by leaving a comment on Brian’s blog and/or tweeting at Brian and Jenny.

Another signficant step forward announced was a test verion of a ‘CERIFy’d’ Research Outputs System (ROS) which had just been made available on the first morning of the workshop. A demo can be viewed showing CERIF import. Live use is planned within several weeks. With NERC taking the decision to move to ROS, there will shortly be five Research Councils using the system. Interestingly, ROS plans to harvest from institutional repositories, which will avoid PIs having to submit individual outputs. ROS  staff are working closely with the JISC CERIF in Action project and there are also close parallels with the IRIOS2 project.

The CERIF-based Gateway to Research (GtR) was another focus of discussion. Whereas ROS will be used for institutional input, GtR will be for access. Since data will be sourced from six different Research Council systems with no common ontology, a data dictionary will need to be developed. The project has been advised (by Jimmy Wales of Wikipedia) to concentrate on making the data available in a standard format (CERIF) and not to worry about the interface – instead prize challenges will be offered to communities for developing applications.

As might be expected the issue of identifiers arose a number times, researcher identifiers in particular, with ORCID being recommended by the JISC Task and Finish Group. HESA also highlighted the ‘big opportunities’ for person identifiers.  The prospect of a ‘UK ORCID’ was discussed, alongside the business case and data security issues. JISC will be looking at organisational identifiers next, agreed as a much more difficult nut to crack.

Repository infrastructure development work was presented by the RepNet project at EDINA (aiming to increase the cost effectiveness of open access repositories) and RIOXX (metadata guidelines for repository managers specifically).

A range of breakout groups covered topics including impact, vocabularies/ontologies, institutional repository/CRIS challenges, research data, and options for maintaining CERIF outputs from JISC project (eg role of euroCRIS and CERIF task group). The REF breakout discussion resulted in agreement with HEFCE to develop a CERIF XML template for research groups, staff and outputs submission and to initiate a test pilot for submission (with KCL and the University of Bath – both to be approached). A test pilot will allow valuable learning within a proper framework – import/export of CERIF XML is planned to start in September 2012.

Presentations from the workshop and breakout outputs are available via the programme page. A fuller event report will be published in the next issue of Ariadne.

Slides from the plenary talks and several of the parallel sessions are available on the IWMW 2012 web site.  In addition video recordings of the plenary talks together with several brief video interviews which were recorded at the event are also available on the UKOLN Vimeo account.

The Open Knowledge Foundation’s Open Biblio Group, DevCSI and the

Working Group on Open Data in Cultural Heritage, present BiblioHack, an
open Hackathon to kick-start the summer months:

http://devcsi.ukoln.ac.uk/2012/05/10/bibliohack/

From Wednesday 13 – Thursday 14 June 2012, we’ll be meeting at Queen
Mary, University of London, East London, and any budding hackers are
welcome, along with anyone interested in opening up metadata and the
open cause. This free event aims to bring together software developers,
project managers, librarians and experts in the area of Open
Bibliographic Data.

A workshop will run alongside the coding activity on 13 June, and a
meet-up on the evening of 12 June is open to all, whether you’re
attending the Hackathon or not.

What is BiblioHack?
BiblioHack will be two days of hacking and sharing ideas about open
bibliographic metadata. There will be opportunities to hack on open
bibliographic datasets and experiment with new prototypes and tools. The
focus will be on building things and improving existing systems that
enable people and institutions to get the most out of bibliographic data.

If you’re not a coder, there are sessions for you too. We will be
running a hands-on workshop addressing the technical aspects of opening
up cultural heritage data as well as looking at the best-of-breed
open-source tools to achieve that. There will also be an opportunity to
prepare your data for a hackathon and a look at the best standards for
storing and exposing your data to make it more easily reusable.

For further information on speakers, venue, arrangements and free
registration, please visit the DevCSI Web site

The event report has now been published on the DevCSI blog, some highlights include:

- Considering the use of SWORD 2 and BitTorrent to deal with large datasets
- Proof-of-concept centralised service for tracking activity data around research projects and individual datasets
- User perspectives on metadata for datasets and examining a common schema to describe metadata for datasets

There are several multimedia interviews with attendees who talked about what they learned, discussed and worked on at the event.

We hope you find the report useful and watch out for similar future follow-up events in this area from DevCSI and JISC.

In conjunction with Open Repositories 2012, Edinburgh, Scotland

Text mining tools and technologies have a long history in the repository world, where they have been applied successfully for a variety of purposes. These vary from pragmatic aims such as enabling document search and browse facilities, linking related documents, identifying copies or facilitating the deposit process, to support tools for academic research. The latter category includes supporting research on the basis of a large body of documents, facilitating access to and reuse of existing work, and connecting the formal academic world with areas such as the traditional and social media. The JISC have funded a number of projects and initiatives in both areas, notably NaCTeM and the ResDis programme. Research areas as diverse as biology, chemistry, sociology and criminology have seen effective use made of text mining technologies.

However, the uptake and hence the impact of these tools has been uneven. Several obstacles to development and deployment are frequently cited, including the maturity, complexity, and in some instances cost of software packages, as well as scarcity of relevant technical skills. Text mining methods and tools can be fragile and complex, requiring significant set-up time and effort. Projects making use of text mining may also suffer from legal obstacles, such as copyright and intellectual property considerations. The benefit to be gained from deployment of text-mining tools in areas such as institutional repositories or as a research tool in its own right may be difficult to predict without a costly pilot project.

This workshop is intended to bring together contributions from practitioners and researchers in fields connected to text mining and analysis. Authors are invited to submit original, unpublished research papers: as a workshop, both work-in-progress and completed work are welcome.

This event will take place during the OR–2012 pre-conference workshop session (9th–10th July 2012).

Topics of interest include, but are not limited to:

1. Discipline-specific research involving text-mining: bioinformatics, chemistry, the social sciences, etc.
2. Techniques in text mining: sentiment analysis/subjectivity analysis, opinion mining, affect analysis, metaphor analysis, etc.
3. Legal aspects of text mining/analysis.
4. Current developments in text mining.
5. Metadata extraction from document text, including formal and informal metadata: document indexing, document classification, and evaluation of metadata quality.
6. Text mining for document categorisation or summarisation.
7. Text mining over the social web: community detection, timelines, etc.
8. Evaluation of text mining tools, open-source or commercial: case studies and findings.

Types of contribution

The following possible contributions are sought:
1. Full papers (6–8 pages)
2. Extended abstracts for oral presentation, posters or software demos (1–2 pages)

Submission

Papers/extended abstracts should be prepared in either Word or LaTeX using the Springer LNCS format (http://www.springer.com/computer/lncs?SGWID=0–164–6–793341–0).
Files should be submitted by email to Emma Tonkin

Important Dates/Deadlines

• 15-May–2012 Title/Abstract submission (optional)
• 25-May–2012 Full paper/Extended abstract submission
• 8-June–2012 Decisions announced
• 25-June–2012 Submission of final papers
• 9/10-July–2012 Workshop

Publication

All accepted contributions will be published in the workshop proceedings. Authors of selected contributions will be invited to submit an extended and revised version for formal publication; to this end, a call for chapters will be launched following the workshop.

Workshop chairs

• Emma Tonkin – Innovation Support Centre, UKOLN, University of Bath, UK
• Torsten Reimer – JISC
• Paul Walk – Innovation Support Centre, UKOLN, University of Bath, UK

We have reverted back to the three-day format for this year’s event and since we’ll be starting on the opening morning (rather than after lunch) we are able to provide a fuller programme than usual, with 11 plenary talks and 20 parallel sessions.  Further details about the plenary talks is given in a post on the UK Web Focus blog.

Brigitte Jörg’s in-depth tutorial covered the CERIF data model (eg there are lots of entities but all are described in exactly the same way) and how it is used, including new items in CERIF 1.3. CERIF also has a triple structure, so can link to the semantic web.

The data surgery in the afternoon focused firstly on a synthesis of UK CERIF mapping work, so involved a number of JISC projects, including CERIF in Action, IRIOS (1 and 2) and CERIFy. There was lively discussion on some of the issues to be resolved eg multiple identifiers, person names, titles, vocabularies. The CERIF in Action project is tasked with implementing the resulting model.

Other sessions covered CERIF use in the Research Management and Administration System (RMAS) – institutions will be able to start buying from framework at the end of February; CERIF and research datasets; linked open data; and a taxonomy of research equipment for inclusion in CERIF. A fuller write-up will follow.

Most of the presentations are already on the UKOLN web site. A number of participants used the #cerifbath Twitter hashtag at the event and a record of the tweets has been published on Storify. Twitter users (and, indeed, non Twitter users) at the event may also wish to add their details on the CERIF tutorial and UK data surgery Lanyrd page which will enable you to make links with others with an interest in this area.

Also watch this space for JISC RIM project synthesis posts, coming up soon…

The call for submissions for this year’s event is now open. As described in an accompanying blog post in addition to proposals for plenary talks and 90 minute workshop sessions we also invite other ideas for ways of engaging with the IWMW 2012 participants.

For the first time euroCRIS task group meetings will be co-located, and preceded by a CERIF tutorial and data surgery. UKOLN, euroCRIS and JISC are jointly organising these events on 9-10 February. CERIF, the Common European Research Information Format, is now being widely used in UK higher education institutions as a result of many recent procurements of CERIF-based CRIS (Current Research Information Systems) – particularly during 2010 and 2011. However staff in many institutions have expressed the desire to strengthen their understanding of CERIF in order to better exploit the opportunities offered by this powerful (and complex) standard. The meetings are being supported by the JISC Research Information Management (RIM) Programme which is working to increase engagement with CERIF and support the emerging community of practice in the UK.

An introductory CERIF tutorial will be held on the morning of 9 February. In the afternoon there will be a ‘data surgery’ session which will examine the use of CERIF in real life scenarios. The aim is both to support new adopters of CERIF and to consolidate existing expertise. The afternoon session will therefore be divided into two groups – one for those new or relatively new to CERIF, and another for more advanced users. Participants are being asked to bring CERIF queries and data modelling/mapping issues for discussion with CERIF experts.

On 10 February five euroCRIS task groups will meet in parallel sessions. See the agendas and note that CERIF expertise is not required to participate. Task group leaders are keen to gather UK RIM experience.

The meetings are open, and all those involved in research information management (eg CRIS managers and IT staff, metadata/repository managers, research office staff…) are strongly encouraged to participate.

Note that the closing date for registration is 3 February.

Simon Hodson, the JISC Programme Manager for the Managing Research Data programme had invited me to run a workshop session on “Blogging Practices To Support Project Work“. This was felt to be an important topic for the funded projects as there is a contractual requirement for projects to provide blogs as part of their engagement and dissemination activities.

Running an interactive workshop session for an audience of over 60 people can be quite challenging, especially as many of the attendees were new to me and I was unsure of their expertise and interests and whether they be willing to engage on open discussions.  I had therefore prepared a large number of slides which I would be able to use to cover a variety of topics related to the use of blogs. Despite only a small number of people being activebloggers I was pleased to find the the audience was willing to engage in discussions which meant that I was able to allow time for the discussions to take place, with the slides being a resource (available on Slideshare and on the UKOLN Web site and embedded below) which could be accessed afterwards.

In addition note that a Storify summary of the session, based on the tweets posted during the workshop session are available.

The main points I made during the sessions are listed below:

• Blogs have a role to play in engaging with a project’s user communities and supporting dissemination activities.
• Not every will be a natural blogger (or coder or writer of peer-reviewed papers). However since project must publish blog posts there will be a need to enable those who enjoy blogging to do so and encourage those who may be reluctant bloggersto develop their skills in use of this medium.
• Since many blog platforms provide mobile-friendly access to resources, we might expect access to content hosted on blogs to increase as use of mobile devoices grow.
• Blog platforms can provide machine-friendly access to resources through use of RSS. In order to ensure that full content of blog posts is made available, blogs should be configured to deliver the full content of the posts and not just a short summary.
• Spam comment can be a problem, but spam filters can trap much automated spam.  It is probably better to accept the need to manage spam that to impose barriers which can inhibit interested parties in adding comments to posts.
• There will be a need to manage blogs which the project is completed.

The slides which are available on Slideshare are also embedded below.

This seminar will focus on using archived web sites and collections of web archives for scholarly research.

A presentation on the current state of play will be given, followed by a brainstorming session on research uses of the UK Web Archive. This will address issues such as:

• What types of research questions could be answered (part or whole) by the UK Web Archive?
• Which disciplines do you think would/could make most use of the UK Web Archive and why?
• What tools or analytical techniques should we be considering?
• How might the research opportunities of a selective web archive differ from those of a domain-level web archive?

Who Should Attend?

This seminar is open to UKOLN staff and members of the University of Bath with an interest in research uses of Web archives. If places are available, others with interests in this area will also be able to attend.

Online Booking for Event

An Eventbrite online booking form has been set up for those wishing to attend this seminar.

Biographical Details

Maureen Pennock is the Engagement & Liaison Manager for the UK Web Archive, based at the British Library. She joined the British Library in 2008 as preservation project manager for the Web Archive and, after a brief stint as a digital curator, returned to the web archiving team earlier this year. She has a long career in digital preservation and curation, having previously worked at UKOLN for the Digital Curation Centre (2005 – 2008) and in the Netherlands (2001 – 2005) on earlier digital preservation projects. She has spoken at numerous conferences and published widely on several different aspects of digital preservation and curation, ranging from authenticity and lifecycle management, to email preservation and archiving websites & blogs. She can be contacted at maureen.pennock@bl.uk and is @mopennock on Twitter.

Ed Pinsent is an archivist and records manager based at the University of London Computer Centre with many years’ experience in digital preservation and web archiving. He has been collecting archive copies of JISC project websites since 2005, via the UK Web Archive; as part of this he has dealt with issues of rights clearance and crawler management. Ed also contributed substantially to the JISC Preservation of Web Resources Handbook, managed the AIDA (Assessing Institutional Digital Assets) project and contributed to the CARDIO project on managing research data. Ed also teaches several modules for the Digital Preservation Training Programme, and in 2011 has done consultancy work for the Parliamentary Archives, the Women’s Library and others.