• GovTech Lab

#CloseUp: ŽEIT x Codami

Welcome back to one more interview with GovTech Challenge Series 2.0 participants!

Today we are talking with Aliona Gaidarovič from The Office of Inspector of Journalist Ethics (ŽEIT) who has submitted a challenge for detecting and tracking online hate speech. Our second interviewee is Dominykas Rentelis, CTO at Codami Technologies - they have created a solution that tracks the hate speech in one of the oldest and most complicated languages in the world - Lithuanian. Let's read on!


Tell us more about your submitted challenge for GovTech Challenge Series 2.0 and why is it important to solve it?


Aliona: Our challenge is about creating a one-stop solution (a tool) to fight hateful speech on the internet (social or news media). The tool is supposed to scrub the internet and find comments which might be construed as hateful speech, giving our institution an irreplaceable tool to efficiently verify and block the spread of hateful comments.

The main problem we have at the moment is detecting hate speech, because of the vastness of the media we are monitoring. The tool we use at the moment scrubs the internet based only on one parameter – keywords that are usually connected to hate speech.

What we expect the new tool to provide is automatisation of classification of comments based on contemporary technologies such as machine learning-based linguistic understanding of the comments. This advancement should considerably reduce the requirements of human resources because comments would be categorised, by their likeliness of being categorised as hateful.


Why is it important to look for innovative solutions that could be applied for the public sector use?


Aliona: First, its human workload reduction. Second, it is much easier to effectively evaluate a small pool of results, rather than manually read thousands of comments.


Do you have any tips for other institutions who want to work with startups and apply innovations?


Aliona: We could only recommend having your attitude adjusted to the realities and potential of such "young" technologies (like in our case – machine learning).

The technological approach usually is not very fruitful at the beginning. Complex problems require complex solutions, and the understanding of the problem (the concept of the correct approach to the problem) usually becomes obvious only well into the process of creating such a tool. Our tip would be – patience and trust in the experts.

Complex problems require complex solutions, and the understanding of the problem (the concept of the correct approach to the problem) usually becomes obvious only well into the process of creating such a tool. Our tip would be – patience and trust in the experts.

Tell us more about your submitted solution for GovTech Challenge Series 2.0 and how could it solve a social or public sector problem?

Dominykas: The hate speech automated detection and identification system is based on AI and specialised for Lithuanian language. The system identifies, classifies, and reports if the hate speech appears on the web. This solution will help reduce manual work for searching of hate speech online and will enable allocate resources more for dealing with this problem.

Why is it important to create innovative solutions that could be applied in the public sector?

Dominykas: Because Governmental sector has the great plans to solve big things for all of us in Lithuania. But sometimes institutions are simply lacking of knowledge how technological solutions could let them do the same tasks in quicker and better way. So if you know how to automate some manual work for them, it could be a beautiful collaboration and contribution to our all well-being.

Sometimes institutions are simply lacking of knowledge how technological solutions could let them do the same tasks in quicker and better way.

Do you have any tips for other creators who want to work with the public sector?

Dominykas: If you have needed competences and can dedicate some time for the new projects, just apply for the next GovTech challenge!

12 views0 comments