Novice Area Mission — Liberate the Social Caseworkers with AI
Caseworkers should make more meaningful conversations instead of being trapped by administrative work.
Hey you, come on in! Welcome to Omdema Legend Land. In this stage, you will help out a 96 years old NGO to be geared to the digitalized world by building up a machine learning project. Are you ready?
Find joystick and choose the lane
In every game, a novice area is always a place where you can be armored after completing simple missions. The first step is to learn how to use the joystick which is the controller for the entire project and which sector is the one you’d like to put more effort on and you’re more interested in. If you are not sure which way to go, then look for power leveling from senior engineers and follow up with the walkthrough.
Strengthen skills and unlock new ones
With only one basic attack, you won’t have enough time to kill the mob. This is also what I have learned from the project. As we need to deal with two main issues the social workers are facing :
- Endless and inefficient document checking process
- Language barriers with “children on the move” service
In a word, knowing how to build models is not the end solution for the ones listed above. After the brainstorm, the team has come up with constructing a web app as an output to retain the build-up models in a more visualized way.
Clear a jungle and destroy towers
Often times, newbies are so excited for the big boss and forget about their position — next to the jumping-off point unarmoured. However, within the project, the seniors showed me how to split tasks in a measurable way.
- How to help the caseworkers to prioritize the urgent cases without checking every single detail of the conversions and records?
We have been displayed different approaches like manually scoring each case with the risk level in order to predict the priority. Caseworkers can make their own top list based on the score. The score comes from the similarity calculation and key verbs linked with abuse type.
- How to debrief each case from different receiving formats from emails to pdf files?
We have summarized cases by extracting the key messages and background of the caretakers in a tabularized format to clearly reveal the storyline.
Work as a game player, not an NPC (non-player character)
Once you’re in the game, be active in all discussions even though you can’t contribute something essential but it can help you to be in the team and understand better how other data scientists are thinking and working. You can start with a small mission like data scraping, named entity recognition (NER), etc.
Group up for the team fights, don't be a solo
In a Multiplayer online battle arena (MOBA), communicating with your team can earn you more rewards and level up with fast speed. When it comes to a machine learning project, it comes to the same. Even with the task distribution, sometimes you may feel confused. Don’t hesitate to ask for help from task managers in this case and get through the task calls when possible.
When you catch yourself in a crossfire, step out and take it as a game. You will win the reward with a good strategy and level up in the team.
