- With daily changes to the COVID-19 situation in Canada and updates primarily being released in English and French, it’s especially challenging for individuals to stay up-to-date on the COVID-19 situation if they are not fully comfortable in English.
- Government updates on the COVID-19 situation are available first in English and/or French and additional translations often come later. Community hubs struggle to send volunteer-translated updates to their communities and there is no centralized source for reliable, up-to-date, and accessible translations of the latest COVID-19 news. We found that the Alberta public health site has information available on the latest COVID-19 variants in their English fact sheet but not in their Spanish fact sheet, for example.
- This could be a useful tool for international students who would be more comfortable receiving critical news updates in their native language.
- While we were working on this project, a newcomer made a callout on social media to ask where to find the latest COVID-19 updates in Ontario.
- The app could also feature as a language learning aid for recent immigrants.
What it does
- The main News Feed is updated daily by an API which queries worldwide news stories for headlines and/or articles which contain the keywords ‘Ontario’ and ‘Covid’.
- The Google API automatically provides a translation for the news story when there is no community member translated version of the information.
- Community members can then edit and verify the accuracy of the translation provided by the Google API, increasing credibility and trustworthiness of the news stories.
- Bring community members together like old times with a news having multiple dimensions ( same with the translating part ) every translation leaves a touch of the translator and their community.
- Scanning the QR code on the article page with a mobile phone redirects the user to an AR Host on their device, which then reads the news article aloud in the user’s browser language. This provides an option for visually impaired users or people on-the-go.
How we built it
- Designs: Figma
- Frontend: React, Redux, CSS3
- Backend: NodeJS, Express, REST APIs, Google Translate API, Google Translate AutoML API, and News API
- All the data is stored and queried using CockroachDB
Challenges we ran into
- Identifying free (or affordable), open-source APIs to perform the required tasks for the application.
- Using the free version of the news feed api ( we can only get brief article not the full article )
- We had difficulty finding translated COVID-19 articles on government sites. We were able to train a translation model using Spanish & English COVID-19 updates from Public Health Ontario, Peel Region, Hamilton, Vancouver and Alberta. Although we were able to generate 100 translation pairs in .tmv format, an additional 100 translations would have been needed to complete the model. So, we instead created the placeholder code for the AutoML API that could be used when an initial model is complete, and implemented the Google Translate API for the hackathon.
Accomplishments that we're proud of
- Developing a MVP from idea to design to development within the time frame.
- Using the Translate API for the very first time and implementing it was a huge success.
- Integrating REST APIs and testing with Postman/Insomnia
- Setting environment variables and preventing our API keys from being uploaded to github
- Great communication between team members, timed meetings and keeping track of each other’s parts using Trello, allowing us to divide the work properly to accomplish a working solution.
What we learned
- How to work with people who are in literally 3 different time zones.
- How to incorporate various APIs into one application.
- How to collaboratively plan ahead how to integrate different components before building.
- How to train a machine learning model and what constitutes good training data.
What's next for CCN
- Making a leaderboard for all the translations done which will bring a sense of challenge and within different communities members.
- In tandem with the leaderboard, implement a karma system to encourage more community members to collaborate.
- Allow users to add their translations not only to our database but also to train our machine learning model.
- While developing the app, we could not find any translation API that provides translations for Canada’s Indigeneous languages and we could not find a machine learning service that would allow us to train a machining learning model for a language that was not already supported. In a future version of the app, we would like to allow users to add translations to our database for languages that are not available with typical translator tools.
- The query parameters could be altered on a by-province basis so that it would service users in all provinces and territories.
- Following from above, users could search for any topic that they want to read about.
Check out our Github repos
Prize Categories for CNN
- Deloitte One : making a environment where language barrier is not an option, each and every immigrant can be updated about the latest beings with the community, a neutral news platform run by the community.
- Sun Life Challenge : we are making a platform where virtual teams all over the world can help the team member who is around the globe to engage with the latest happening in a language which is easily interpreted by the person
- Echo AR
- Use of Google Cloud