[Student Essay] AIteen Academy - Sharing Love… through AI!

Home > Think English > Readers' Opinion

print dictionary print

[Student Essay] AIteen Academy - Sharing Love… through AI!

Grade 11, Minjoo Kim

Grade 11, Minjoo Kim

 
Grade 11, Minyoung Kim

Grade 11, Minyoung Kim

 
There is nothing more you can do but study. Keep up with our expectations. We feel a sense of responsibility to our family, teachers, and friends - to be the perfect daughter, a kind friend, and a model student. But trying to show the best version of ourselves led us to a wall that blocks our path. To overcome this we need support. However, the process is slow and takes effort and time to find the courage to talk about this issue.
We were the lucky ones. We were loved and supported throughout our difficult times. Now we want to share love and support with others. After several discussions, we came up with a plan to develop a survey and a machine learning program that will detect teenage chronic stress and recommend possible ways to alleviate it through recommending songs. We divided the whole process into several steps.
 
Step 0. Seeking expertise  
It was not that simple to start. First, since we were dealing with sensitive issues, we had to consider the ethical implications of our activities. We contacted local expert organizations, including the Korea Youth Counseling and Welfare Institute, via email for data or possible collaborations in constructing our alleviation system. Despite our efforts, it was difficult to get a definitive answer as a school club that didn’t even accomplish anything yet. So, we decided to take advantage of the many resources at our school, including our social counselors as well as our supervisor, Dr. McAdams.
 
Step 1. Designing a survey for Branksome Hall Asia Students
In order to gather data most relevant to our target audience, we designed a survey specifically for our Branksome Hall Asia middle and senior students. This step was more difficult and complicated than we thought. We put the utmost care into this survey for weeks in organizing questions, making options, constructing carefully nuanced questions, determining a neutral background color, and so on. Even after all this, collecting data was not so easy; however, we managed to gather over 100 responses and proceed to the programming stage.  
 
Step 2. Artificial Intelligence (AI) Ethics club
Understanding the ethical implications of our activity, we decided to allocate time for discussing AI ethics and relevant current issues. We discussed well-known dilemmas entailed with self-driving cars and AI applications in healthcare, using resources from Inspirit AI. We could understand the ethical implications of AI and our project, so we were more cautious and sensitive in constructing the survey.
 
Step 3. Coding our survey and recommendation system
In addition to the gathered data, other data had to be gathered. Since we were recommending music in response to the survey results, we found on Spotify and another application programming interface (API) for the lyrics. Both datasets had to go through a cleaning process, where we either had to fill in missing data or categorize the data. We considered lyrics as one of the most important factors in recommending the most relevant songs as words can trigger feelings that may be harmful. Other factors we considered are explicitness, valence (music positiveness), and energy. Through processing this data, we were able to categorize the songs into random, cheerful, offensive, depressing, sad, relaxing, and neutral. For the survey results, we used a deep learning model to create a vector that would provide information about the responder’s mental well-being level. For the lyrics, we used a pre-trained text embedding model which can map the text into a vector with the same dimension as the survey result vector, so they can be later both compared with each other when someone responds to the survey. Using cosine similarity, which measures the similarity between two vectors, the algorithm is able to sort through the songs to find the top matches with the responder’s results. The system is able to select the top 5 songs with the highest similarity score and return it to the last page of the website.
 
Step 4. Creating a chatbot  
We came up with the idea of creating a chatbot through Kakao while we were working on the recommendation system. Since we have such a precious opportunity to have caring teachers, we thought it would be great to seek help from them. Our chatbot is for students who did not have a chance to seek advice from teachers or are too shy to approach them yet. Students enter ‘student utterances’ and the chatbot returns pre-entered teacher responses that are relevant to the student’s utterances through identifying keywords. Despite the time it took, we were able to gather helpful and heart-touching teacher responses and set up our Kakao account. Then we categorized the anticipated utterances and teacher responses for the Kakao chatbot’s machine learning algorithm.  
 
Step 5. Designing our website  
In order to group all of the projects we had been working on, we decided to design a website that contained everything we did. On the website, we have linked our survey and recommendation system as well as a blog page to post articles about mental health.
 
NOW WHAT?  
Recently, we have just finalized all three of our final products, and we are excited to promote and share our project with our school’s students. All of our teammates have been working together with persistent energy, and through this collaboration, we were able to finish this project meaningfully. Although our algorithm is a small step forward, we learned how to share our thoughts about our emotions and empathize with each other. This project started as a small service club, but now we see that this experience has created a change amongst ourselves. We are anticipating this change to spread out to our friends and local community. Who knows? Our project may bring greater changes, where more friends join this movement of spreading love?  
 
 
Log in to Twitter or Facebook account to connect
with the Korea JoongAng Daily
help-image Social comment?
s
lock icon

To write comments, please log in to one of the accounts.

Standards Board Policy (0/250자)