Please enter your email address to get this file.


Top 3 hacks to a successful semantic Hackathon

CTO, James Malone shares his top 3 hacks to a semantic Hackathon, following SciBite's first US-based Hackathon in Boston, MA.

SciBite Hackathon 2018

Hackathon, as a word, has always struck me as something of a misnomer. The name is a portmanteau of hack (exploratory coding) and marathon (long distance run). But in reality, they are more like hacksprints, or maybe even more accurately hacktriathlon.

Following our 2018 SciBite User Group Meeting, hosted by our friends at Pfizer, we ran our very first US-based Semantic Tech Hackathon in Boston, MA. Our aim was to bring together like-minded individuals working with data or code, including data scientists, software developers/engineers and bioinformaticians, to find creative applications in applying AI to the life sciences.

To aid this, we opened up some of our award-winning semantic text analytics tooling, specifically TERMite (our entity extraction tool) with a large number of VOCABs (the core vocabularies that TERMite uses to annotate terms), TExpress (our pattern matching tool) and Semantikit (our machine learning toolkit). Once everyone got their hands on the starter code we supplied, we handed over the baton to the 31 people that attended from across the life sciences industry and academia.

During the event, one larger group led by Pfizer, tackled the challenge of using OCR to transform hand-written medical text into machine-readable documents, then tag with TERMite, index, and make searchable. They presented their work at the end of the event, which had achieved some remarkable results. Having something working in just one day was quite an achievement.

Another team focused on the use of TERMite annotated data for integration into the ANZO platform as RDF. Again, the team demonstrated the pipeline they had built for extracting, transforming, and loading data and showed how querying would now be possible using the ontology annotations. Many other teams or individuals spend their time exploring the technology, such as mining patterns in patent data and discovering how to enrich vocabularies with Semantikit.

Overall, the day proved invaluable to participants who gave encouragingly positive feedback, but as ever there were take homes we have learnt and will build on for our next Hackathon in 2019 which are worth sharing.

Hack #1. Hold an internal pre-hackathon to create starter code

With only one day for the event (although remarkably a lot was achieved), we decided to help the hackers by running the first mile for them. We held our own internal half-day hackathon during which we wrote some starter code for using our semantic tools and made them available on our open SciBite Labs GitHub repo. This initial legwork proved invaluable as almost everyone used them.

Hack #2. Encourage hack-teams and networking where possible

Organising people at the start of the day by grouping those who wished to work with others by their areas of interest, helped people form teams and encouraged the networking element of the day. Participants could share tricks and tips with each other and make new connections with those from other organisations. About half grouped into teams, whilst the other half largely worked solo. Our hackathon was not intended to be very prescriptive, so this fitted with our expectations.

Hack #3. Set a Hackathon challenge for everyone to take part in

For the Hackathon, we posted several example ‘topics’ that we thought participants might wish to work on in advance of the event, these included ontology learning, document classification and semantic visualization and summarisation. However, some of the feedback we received was that having a single challenge for the event would have helped attendees focus on an end goal. In future, we may present just this – a single challenge with a prize for the winning participant(s). Again, this is not to be overly prescriptive, but more to add a defined objective for those that wish to pursue one.

Having the chance to play with new technology can often be difficult during normal, busy working weeks, so having a day away from the office to try new things in a room full of like-minded hackers is always a valuable experience. At the end of the day it was great to see everyone’s enthusiasm when presenting their hard day’s work, as well as seeing the new relationships built, but most importantly, seeing that everyone had fun! Let’s not forget to mention that everyone left the day filled with pizza and beverages (including beer)…

Stay tuned for our 2019 Hackathon news.

Related articles

  1. A Hacker’s Guide to Understanding Bio-ontology Jargon

    Perfect for those new to bio-ontologies or who work with ontologists - a whole new vocabulary deciphered!

    Read
  2. SciBite Semantic Tech Hackathon 2018

    As part of the 2018 SciBite User Group Meeting, hosted by our friends at Pfizer, we are also hosting our very first Hackathon.

    When: Friday 12th October 2018 | Where: Pfizer, Kendall Sq, Boston, MA | Time: 09:00-17:00

    Event attendance is by application ONLY and is FREE.

    Read

How could the SciBite semantic platform help you?

Get in touch with us to find out how we can transform your data.

Contact us