From Dev8D

Contents 1 Eprints 3.2 API/DATA Challenge 2 Bounties 2.1 Install EPrints 2.1.1 Start by 2.2 Metadata extraction 2.3 Semantic EPrints 2.3.1 start by 2.4 Disambiguating EPrints

Eprints 3.2 API/DATA Challenge

Speaker Name: Chris Gutteridge/Partick McSweeney

Speaker ID: User:patrickmcsweene

Start Time: Wednesday, 16:30

Eprints API/DATA CHallenges is to create something cool to promote the new features of 3.2.0 especially looking at the use of linked data. There will be four small prizes handed out of Amazon vouchers. 16:45

Session: Expert Talks Wed PM

Bounties

Install EPrints

REWARD: a pack of haribo and your name imortalised on the eprints wiki for all to see. and potentially more for really cool implimentations

Install EPrints 3.2 and write the hello world plugin. Couldnt be easier!

Start by

• reading The eprints installation guide
• then read how to write plugins

Metadata extraction

REWARD: £50 of Amazon monies + a pack of haribo and potentially more for really cool implimentations

EPrints has piles of pdfs in it. One of the biggest challenges we face is automatic extraction of metadata from the text of documents. You can save yourself a lot of pain entering metadata this way. We are looking a tool which extracts text from a pdf and does some clever text parsing on the pdf to extract metadata from a documents "Reference section". Best entries will cleanly cover some of the major citation formats and be able to tell what citation format a document is using. Metadata should include:

• title
• firstname/initial familyname pairs
• publication
• year of publication

additionally you might try and extract

• publication type
• pages it appears on

Semantic EPrints

REWARD: £50 of Amazon monies + a pack of haribo and potentially more for really cool implimentations

EPrints 3.2 (currently rc2) makes tonnes of linked data in N3, N-Triples and RDF+XML. We want to connect eprints.ecs with any external linked data source! Find some interesting information or make a cool visualisation or both. Infact we will take what ever your offering!

start by

• Export a search or a record from eprints.ecs.
• Read a php tutorial about parsing linkdata from eprints http://lemur.ecs.soton.ac.uk/~cjg/Graphite/eprints.php

NOTE: you dont need to know much about EPrints to do this. You certainly dont need to know any perl

Disambiguating EPrints

REWARD: £50 of Amazon monies + a pack of haribo and potentially more for really cool implimentations

write a cool visualisation tool which helps a repository manager decide if two items in the repository are the same or versions of each other. You might want to lay this out as graph! If you are really flash maybe you could do a similar thing with author names. David Williams wrote this paper and D. Williams wrote this other paper. Are David Williams and D. Williams the same? what evidence do you have? They were uploaded by the same person? They are on the same subject? They have both published other papers in similar conferences/journals? They both use the same latex stylesheet?

Good entries have clever mechanisms for working out if things are the same and or a visualisation which makes it easy for a person to tell if two people are the same.