3/4 Chinese - Trendly

Front page
AI summaries
AI summaries
Trend graphs
Latest news
News articles relating to search query

Inspiration

We realized that searching and analyzing the latest news articles can get tedious! Taking ideas from Ground News, we wanted to make something that can easily search through numerous articles, compare ideas and topics (especially in this time), and summarize the true facts. We also included keyword trends and their growth in relevance.

What it does

Trendly is a news analytics tool that can help people find and break down the latest news trends on a topic of their choice. This tool can be very useful for researchers, students doing a project, and just people who want to stay up to date on the most recent news, helping them find and understand the fast-changing world.

How we built it

We built this program on Replit, an online IDE that allowed us to collaborate more easily. One of our team members started with a Typescript template, since that was the codebase format that he was most familiar with. From there, we started working on the frontend, using ideas from previous websites we made. We went for a concise, modern look that streamlines the user interface. Two of our other members got to work on the backend, working to figure out how to use NewsAPI, Gemini AI API, and SerpAPI. We first implemented NewsAPI as our base, using the HTTP GET method to retrieve the article data. From there, our backend programmers wrote code to convert the JSON response into something we could use. We then implemented the Gemini AI summary, where the only difficult part was prompt engineering. Lastly, we wanted to include Google Trends, so we settled for SerpAPI, allowing us to view data that was scraped from the results of the Google Trends search page.

Challenges we ran into

At first, we had trouble coming up with our idea. We tried looking for simple APIs to use and get data and trends from, like Instagram, TikTok, X, etc. However, we quickly realized that those APIs were a little tricky to get our hands on and shifted towards another idea. We wanted to web scrape to find news and media trends, which landed us on our design path.

Accomplishments that we're proud of

We're extremely proud that we were able to accomplish so much in such little time and get 3 different APIs to work almost perfectly together! We also built a system that automatically analyzes news articles and generates trend graphs based on keywords extracted directly from the article content.

What we learned

We learned how to integrate NewsAPI, Gemini AI API, and SerpAPI into a single full-stack application. We also learned how to process and clean unstructured text data, such as article titles and descriptions, by filtering out stop words, validating user inputs, and generating graphs based on fetched data.

What's next for Trendly

We got a lot done in the short time we had; some things could be improved on our website. Firstly is the user interface, it looks great, but we think it could be cleaned up a little more to include fewer news articles. Secondly, we want to have clearer news trends and more in-depth analyses on news topics. Right now, the AI summary on news trends is too concise and vague, and more time can be spent on refining the news analysis. We hope to continue to refine the website and make the experience better and more helpful to users.