AWS Logo
Menu

The Journey of Building a Recommendation Engine App

An application that generates topics and items related to the information users provide.

Published Mar 11, 2024
Project Story: The Journey of Building a Recommendation Engine App
Idea Summary
Life is about making choices, which becomes quite challenging when many options are available. Which beach should I visit for Spring break? Which music should I listen to? Which movies should I watch? Which restaurant should I visit? The "Recommendation Engine of Everything" that I built, based on PartyRock, will make all such recommendations based on a shortlist you provide.
Leveraging the Amazon Bedrock platform, this AI app (the recommendation engine) will prompt users for several simple inputs, including categories (music, chemical elements, movies, food, etc.) and a shortlist of their favorites. Based on these user inputs, the app will provide a list of recommendations, complete with a brief explanation for each choice and pictures of the recommended items.
Inspiration
I started frustrated; most times, suggestions drawn by Googling would be impersonal and irrelevant.
All of this begs the question: Is there a possibility of a more intuitive, personalized way to discover new interests? This idea led to the creation of a Recommendation Engine app that gets under the skin of user preferences and shows recommendations that simply find resonance with users.
What I Learned
There was a path very enlightening from idea to execution: an application is a product developed through creativity and additional technical skills, which are needed in this case to come up with the design of the UI so that it is interesting and to create the user experience so that it is irresistible. Value in sticking through the roadblocks with persistence and how precious value the user feedback holds was taught through user interactions in the real world; these were what planning in theory could not reflect.
How I Built the App
The development process began with identifying common queries people have. For this, an extensive study was done through questionnaires and the analysis of search trends. To cater to this, a core capability of taking user input to analyze the same with machine learning algorithms was designed at the base of the app, which is then matched with a categorized database to derive relevant recommendations. This called for deep dives into both natural language processing and recommendation system design—starting from a minimal viable product and iteratively improving it based on user feedback.
Challenges
Ensuring that the app's recommendations were closely related to user inputs was a significant challenge.
This has required not only understanding the context and intentions of the user with these keywords but also understanding the other big challenge: to be a complete solution for different categories of recommendations and, at the same time, keeping the application interface simple and clear for the user.
This required innovative design solutions and a scalable backend.
Conclusion
The experience of building this Recommendation Engine App was just too fulfilling. It reinstated the fact that there is a need to marry technology with creativity and solve real problems. This project has looked deeply into the worlds of app development and user experience design. With these many challenges, this app represents what happens when you reimagine how it would be to find and explore your interests.
 

Comments