InstaChecker

Inspiration

We were inspired by the IST challenge to reduce the ill effect of social media, specifically misinformation. As college students, we are well acquainted with information in us and many of our peers being provided by social media, primarily instagram. We wanted to build a way for users to easily assess how much of that information is real. Often times, platforms that do similar checking require pausing completely what you're doing and don't provide source for their reasoning. We wanted to address those issues as well.

What it does

It allows you to send an instagram reel to the app and then will identify the likelihood of misinformation of the claims it makes and sources to back this up. It gives you a rough estimate in the form of a percentage of how truthful the claims are and flags reels most likely to be false. You can navigate by reel in the search bar to see old data and click on a tab to zoom into the reel: the risk level, why it was flagged, and sources the app is verifying this from. Along with this, we tried to make the app visually appealing, including the username and photo of the profile the reel is associated with, and so on.

How we built it

We used a sqflite database, flutter app coding for ios and android, and ios and android emulators. We wanted to consider environment constraints, so the app doesn't rely on LLMs and rather uses low-scale similarity and sentiment checkers paired together to get rough estimate of claim validity. We used flutter as an app building software, which utilizes dart code. To store the information about the reels, we used a SQL database that is localized to the device, specificall sqflite. To screen web and instagram results we combed through the html and metadata using flutter extensions. We used ios and android emulators to simulate phone functionality through Visual Studio and Android Studio. All of this was centered around a github connected to our Visual Studio Code. None of what we used costs money.

Challenges we ran into

None of us knew how flutter worked or how to use dart code or emulators. We figured it out by sorting through documentation, coding/downloading software and debugging mistakes, and Gemini input. Setting up each dependency and figuring out how they all worked was probably the hardest part. Almost all of the software was new to us. There were also a lot of different tasks, so we had to figure out team specialties or interests to divvy up the work while still acknowledging how the different code pieces would rely on each other to run.

Accomplishments that we're proud of

We met each task knowing nothing and came out with a coherent demo. We're proud of how the project evolved from a goal, to a design, to a feasibly designed product. We learned a lot and are proud of how we didn't give up face with unexpected challenges.

What we learned

We learned new software, as I have mentioned. Almost everything is new to us in this project. We also learned how different code relies on each other, like how claim validity analysis relies on the html readers or how everything needs to be changed in how it's routed when you implement a databse, among other things. Despite these dependencies, we worked on solo aspect when applicable and came together as a team to stitch the product together.

What's next for InstaChecker

In the future, we might implement this as a widget that flags posts while scrolling and improve its efficiency. As well as this, we could add a persuasion checker that determines if a reel is trying to convince you of something, like an advertisement. We can also implement this across other social media platforms. We will definitely be installing it on our phones to see how it can be implemented in day to day life!

Built With

• android-studio
• dart
• flutter
• flutter-app-coding-for-ios-and-android
• sqflite-database
• visual-studio