• 06 Dec 2017application deadline
  • 09 Dec 2017event
  • 12 Dec 2017short list
  • 15 Dec 2017presentations
  • 18 Dec 2017winners announcement
  • 22 May 2024today


Machine Learning Hackathon

The art of progress lies in the power of extracting structure out of chaos. Strengthen your diligence and get ready to navigate through a realm of unstructured content! Keep in mind to always leave the data better than you found it. All dots turn connecTable on December 9th.

You get a collection of PDF files that are, by design, unstructured documents. Their internal format only contains information about the absolute positioning of elements on the page. The information presented in them is, however, often structured by nature: tables, ordered lists, forms, charts, etc.

You can have data without information, but you can’t have information without data. So, the idea(l) is to enjoy uncovering “all things structured” and provide the ability to convert the resulted data into valuable information.

Extracting valuable structured data from unstructured content turns out to be a difficult challenge. ML techniques are of massive help in this problem. Since we are talking about textual data presented in a visual manner, both NLP and Computer Vision methods can be applied. What is your weapon of mass extraction?

We’ll appreciate the quality of your research with a focus on your vision and the creative angles you approach this challenge.

The best solutions were rewarded with prizes worth a total value of up to 1500 USD.

Hackatune in!

Terms & Conditions


Bachelor’s, Master’s or PhD students
Fresh graduates


The University of Bucharest –
Faculty Of Mathematics And Computer Science
(Strada Academiei nr. 14)


  • 08:30 – 09:00 Welcome! (setup, say “Hi!”)
  • 09:00 – 09:30 Additional info. (theme, rules etc.)
  • 09:30 – 14:00 Focus time! (part one)
  • 14:00 – 14:30 Pick your topping, it’s pizza time!
  • 14:30 – 20:00 Focus time! (part two)
  • 20:00 Time to present your work!


  • Form a team of max. 4 participants.
  • Choose an unique name for your team.
  • Sign up and save the date in your calendar.
  • When the alarm goes off that day, wake up inspired and ready to roll.
  • Bring your laptops. We’ll provide 220 and WiFi.
  • Knowledge is power, but enthusiasm pulls the switch. Let yours be contagious! We’ll bring more details on the topic along with pizza & drinks.
  • Mine. Extract. Connect the dots.
  • Take some time to prepare your presentation.
  • Showcase your results at the end of the day.
  • Win.



Lasso Net of Principled Likelihood Maximizer


Classical Jokers


The Mean Squares


Artificial Intelligent Guys


Brute Force



Codes and Chords


The Hive

CMOS Torino


Next Phase

Since there were so many teams signed up for Textract, we adopted an additional selection phase. Following the Jury’s assessment, the following 6 teams have been selected to thoroughly present Saturday’s work:

Artificial Intelligent Guys

Brute Force

CMOS Torino

Codes and Chords

The Hive

The Mean Squares

All presentations will be held on Friday, 15th of December, at The University of Bucharest – Faculty Of Mathematics And Computer Science (Academiei nr. 14), starting at 18:00.


It’s rewarding to see Textract growing. This year’s edition was substantially more productive than the last. The fact that more and more people are interested in Machine Learning is empowering, as it continues to expand the ML community we’re glad to be a part of. Along with an increase in the number of participants, the overall level of the competition was superior. It was captivating to witness diversity in approaches and we appreciate the quality of the submitted results at the end of the hackathon.

(drum roll) And the winners are:

Brute Force

Prize: 1000 USD gross value.

The Mean Squares

Prize: 500 USD gross value.