Hack the News Datathon Instructions

Dear participants, we highly appreciate your participation in the Hack the News Datathonand we are sure that your contribution will make for another great Datathon. Please, see the participant instructions for the Hack the News Dathaton below. During the event, your task would be to solve the case for Propaganda Identification.

 

21-24.01.2019

Log into DSS platform

Complete your profile for matchmaking up to 22 January

Log into DSS platform

Read the presented case and explore the data set

Login into Data Chat

Speak with mentors and experts about the platform and the cases

24.01.2019

Team forming

25.01.2019

Check out the schedule for the opening 

Register the team

Team forming if you are without a team

Create your own team channel in Data.Chat

Post your team details in an article

 

26.01.2019 – Exploration day

You have the time!

9:00 – 10:00  – Q&A with experts at Data.Chat

27.01.2019 –Finalists day

15:00 – Upload your results and publish the article.

17:00 – Finalist announcement

29.01.2019 – Appraisal day  

17:00 – Finals and closing (subscribe in YouTube and book your ticket)

21-26.01.2018

Register into the Datathon Platform

• Register in Datathon website.
• The link is sent via email.

The registration will go through e-mail confirmation.

Complete your profile for matchmaking up to 22 January

• Fill in your profile details

Make sure that You have filled-in Your profile, including an avatar (not necessary Your actual photo), Your name, Your data science interest, etc.

• Complete your profile if you are alone and need a team.

Log into DSS platform

• A confirmation email about the registration is received. Keep in mind that there might be a delay for receiving it or could be in the spam folder.
• You can directly access the Log In from the page of the Hack the News Datathon.
• In case of issues please write an email to [email protected]

You should enter the platform via this page:

Read the presented case and explore the data set

Read the presented case and start studying the problem domain.

There are some preliminary guidelines to help you with the case.

Explore the data set. (TRAIN SET)

Explore the Tools.

Explore the DEV-input Dataset.

Explore how to access data directly from the data.cloud (Jupyter notebook file .ipynb).ipynb file

******

Keep in mind that the DEV SET will be opened on Friday with the official opening of the Datathon (follow the #announcement channel in the data.chat). You will be able to test your solutions with the TEST SET that will be shared on Saturday evening!

* All of your questions regarding the data set you can ask in #nlp_text_mining channel.

Login into Data Chat

Once again use the link from the top panel to access the Data.Chat or enter in your browser https://ask.datasciencesociety.net

Now comes the fun part! We highly recommend being part of the conversation. One of the most valuable aspects of the Datathon is to be part of the online community. For global communication with the participants, we use DataChat among other platforms.

Join the Data.Chat and find your teammates there.

• The Data.Chat is the main place for online communication among the participants, experts and mentors. If you’re familiar with Slack, this one works very similarly. It is in essence Rocket Chat (help: https://rocket.chat/docs/user-guides/; App Store: / Google Play) and it also has many cool and free features like Videochat, full customization, etc.
• If you want to install the Rocket Chat application on your mobile, you must be logged into the DSS platform first.
• All the communications are conducted in English.
• If you have any technical problems write in the #issues channel

Speak with mentors and experts about the platform and the cases

There are several groups which we can recommend to join, use the search menu to find them:

• #general – for general info
• #announcements – read-only group for announcements from DSS
• #issues – if there are problems also you can check Help -> FAQ menu on DSS site
• #nlp_text_mining – this is the channel where the case is going to be discussed with experts and among contestants
• #ml, #sna, #cv, #ai, #bio_inf – other interesting areas of our platform
• and the most important, you are free to set up your team group and start working in a private channel

These channels are going to be the main place of where you will ask questions to the mentors during the Datathon.

If you are the leader of your group – creating the channel for your team is easy, just click on the sign (marked with red) next to the search.

To invite people to your channel write /invite @name and you will get auto-suggestions.

Video and audio chat is conducted via direct messages in the group itself.

If you are a mentor or a society mentor let @denitsa know so he can add you into operational group #mentors-internal

24.01.2019

Team forming if you are without a team

• Take part in the Matchmaking process on Thursday night where you can see what DSS recommends to you.

• • Matchmaking requested applicant from their profile can see our suggestions for a teammate in #announcement channel.
  • If someone is left out during the teaming up in the #announcement channel he/she should move to #general to pair up with other lonely participants and be proactive.

Team forming

Join the recommended channels and talk with the other participants, get to know them and try to form a team with them if you already don’t have one.

• The preferable size of the team between 3 to 5 people

 

25.01.2019

Check out the schedule for the opening

https://www.datasciencesociety.net/datathon/#schedule

• The updated schedule is on the Hack the News Datathon page. The Datathon starts officially at the same time Worldwide at a live-streamed opening ceremony. There would be some clarifications and possibly other useful messages given by DSS.
• Click on the locations Bulgaria, India, Qatar and Saudi Arabia  for local schedules.
• Keep in mind that the times are in UTC (+/- 00:00). If you are participating physically, you should check the local hosts’ articles for the local schedules which are published in News and as a reference to the Datathon page on the Local Hosts’ cities.
• It is important to follow up the opening schedule so you can find out what is coming up immediately.
• There will be live streaming with updates in the general Data Chat.

* Please show up at the local Datathon Hub on time on the starting day of the Datathon. You will use your own tools, including hardware, software, communications, and stationery. If you need a computer and certain software, make sure you have it prepared.

Register the team

Register your team until 19:00PM (UTC +/- 00:00).

Start working on the team case solution:

Create your own team channel in Data.Chat

• • • Find each other in direct messages and appoint one of you to create the team private channel, with the most strange or funny name of your team if possible use animals
    • Select icon with the red mark, next to the search box.

Post your team details in an article

• • • You need to create a new article and add all teammates plus team name, case,  and mentors. (the initiator of the article doesn’t need to include his name himself in the tags)
    • To create article follow these steps: from the top menu click My Profile ->then My Articles (on the gray line) -> then click New Article. The title should be Datathon – HackNews – Solution – *the cool name of your team*
    • There is a template one which can be used called “Example of Team Article”.
    • The role of the mentor is to read and comment on the article (your result) without editing it, while it is in private mode (Draft status).
    • On the top of the article include your name, registered nickname, and email.

 

26.01.2019 – Exploration day

You have the time:

• • • Contact mentors in Data.Chat for clarification and mentorship, if needed set up audio or video call.
  • Normally around Saturday morning, there are many questions by the participants of the Datathon. That is why we organize live Q&A session on Saturday morning in which the industry experts and the mentors provide some clarifications about the cases and about the approaches to solve them. Your attention would be very beneficial. There will be live streaming with updates in the general Data Chat.
    • There is an #issue channel for DSS platform problems which can arise
    • Have fun at the Datathon, make videos and share moments in the social media using #Datathon and #HackNews hashtags

For sharing your stories, we recommend you to use Twitter(@datathons) and Facebook(@datascisoc) where we want to capture the best moments of the Datathon. We invite every team to post a short team video on Saturday (up to a minute) and a picture on the social media adding #Datathon and #HackNews (and other official hashtags given by DSS). Let’ s create a story and tag ALL pictures and videos so the world can see our work towards fighting the problem with propaganda in news!

18:00  – The test set is uploaded

You will have the time and the data to test your models before the final submission on Sunday at 15:00.

27.01.2019 – Finalists’ day:

15:00  – Upload your results and publish the article.

• • • Finally, it cаme the time to be proud of yourselves what you did and how much you learn. We advise you to use the CRISP-DM methodology for your masterpiece
    • You need to add export from your code in Jupyter Notebook or something similar and your results. (docx, LaTeX code)
    • The main idea is that the mentor and all the jury will review the results and provide guidance, comments for evolving the solution and improving your knowledge and expertise.

17:00 – Finalist announcement – Jury

• • • This is also the time for voting for everyone else’s solution (by clicking on the upvote arrow at the solution). At the end of the competition, these votes would count for the Audience award.
    • Every group would be judged by an automated scorer for each of the three tasks.
    • The team which gets ranked on the leaderboard at the top of their task qualifies for the final.
    • Participants are also encouraged to leave comments and questions to other articles when they are available. You can easily compare different or similar approaches.

The finalists’ teams are going to be announced and will have time to prepare for the final round on Tuesday, 29.01. They should send their recorded presentation (max. 5 minutes) to the organizers on Monday (28.01 before 18:00 UTC 00:00).

 

29.01.2019 – Winners’ day:

17:00 – 18:30 – Finals and closing

18:30 – 19:00 – Finalists Jury and a LIVE Panel Discussion with the Finalists

• Teams should log in and participate at a live session with questions from the audience and the jury.
• The jury vote will be based mainly on the content of the articles of the team, helped by the various scores.
• The team with the best collective rank wins.

19:00 – Winners announcement

• Awarding the winners

19:15 – Local Data.Celebration Party

There will be streaming for the award ceremony so keep an eye on the website and register to get a streaming link, schedule and the #announcement chat for pinned messages.

---

Article creation and Platform

1. Get to know the case and the DSS platform. (in the linked article you will find links with Video and Case description for the data case)
2. Use our Data.Cloud if you want to use computational power
3. Watch the User Guidelines Manual – How to create Article and Group Chat
4. If you have a team, start creating your team article and tag your members by Friday
5. DL for creating your team/personal article until Friday 19:00 (UTC 00)

IMPORTANT: You must do these 2 steps when the Datathon officially starts on 25th. (these two actions will unlock during the Datathon)

1. Link your article with the working data case. (from Article edit, the drop-down menu)

2. Add your mentors (preslav, alberto giovanni, laura, vsenderov) in the field Mentor in your article.  (Article edit)