Springer Nature SciGraph

A Linked Open Data platform for the scholarly domain

First Springer Nature SciGraph Hack Day "Linked Open Data in Action"

On June 23rd 2017, we hosted our first ever Springer Nature SciGraph Hack Day "Linked Open Data in Action" in London. And what can we say? It was a very successful first SN SciGraph Hack Day, and we are convinced that this was just the prelude for many similar events to come.

Why did we host a SN SciGraph Hack Day?

Our motivation was to engage with the Linked Data researcher community, encourage developers to build cool tools with our data offering and position ourselves as Open Data research publisher. We also wanted to gather first-hand feedback from users of our data and bring together developers from different countries, acting as a hub.

What did you miss?

About 30 internal and external participants came together from different parts of Europe: we had hackers from the UK, Germany, Denmark and even Romania. A lot of different companies and institutes were involved and created a very productive working environment throughout the day.

In preparation for the event we had created suggestions of what do to with our Linked Open Data offering (Data Visualization, Information Retrieval, Linked Data Browsing and Predictive Analytics) which attendees used as a basis to brainstorm ideas together. Some also came with their own ideas.

At the end of this intense day, we had interesting presentations and demos covering the achievements of participants during the SN SciGraph Hack Day before heading out for a well-deserved drink at a nearby pub together.

Download: Results presentation

(PDF, 2.59 MB)

More information about SciGraph

Any questions? Please contact us.

Watch our video summary

What had been worked on?

Our attendees focused on recommendation systems, trend(setter) detection, information retrieval, Linked Data integration, data visualization in the limited time (6 hours) they had.
See some of the amazing results below:

Helping you get into the right journal

New Content Item Problem: Thousands of possible journals, which one matches my content best?
Using a text classifier trained on 5M abstracts using FastText, this service classifies your abstract by journal (using 3500 possible journals).
Just having a journal name might not help enough, so when you click on the journal name it’ll use the triplestore to look up people at your institute who have published in that journal before so you can talk to them.

Uncovering the Hidden in SciGraph

New Content Item Understanding large knowledge graphs such as SN SciGraph is challenging. It’s not that a properly written query can’t retrieve valuable information. The essential problem is to get an idea of the queries that deliver real insight. Those queries are hard to express when the underlying data or schema is not well known.

The video shows a sample exploration of SN SciGraph data with SemSpect. To find the top journals and grants related to articles about cancer, it starts with the subjects, restricts them to those whose description contains ‘cancer’ in the tabular view and expand this group to journals and grants via articles.

Peer Reviewer assignment system

New Content Item The idea was to construct a collaboration graph based on SciGraph

o Nodes are people and organizations

o People are connected to other people if they have authored a paper together

o People are connected to organizations if they have an affiliation

We can now assist the peer reviewing process

o Discover simple conflicts of interest (previous collaborations resulting in publications or shared affiliation)

o Suggest reviewers based on distance in the graph and shared research

Image Gallery

About Springer Nature SciGraph

Springer Nature SciGraph collates information from across the research landscape, such as funders, research projects and grants, conferences, affiliations, and publications. Currently the knowledge graph contains 155 million facts (triples) about objects of interest to the scholarly domain. Additional data, such as citations, patents, clinical trials, and usage numbers will follow in stages, so that by the end of 2017 Springer Nature
SciGraph will grow to more than one billion triples. The vast majority of these datasets will be freely accessible and provided in a way to enable experts to analyze downloaded datasets using their own machines, or via exploration tools on the Springer Nature SciGraph website which is being continuously enhanced.