This project is an AI-driven alert system that monitors Harvard School of Engineering and Applied Sciences (SEAS) public calendars and notifies users about events relevant to their interests.
Unlike traditional keyword-based alerts, this system identifies events that are conceptually related to a topic. This allows users—especially journalists—to surface meaningful events even when the event description does not contain obvious keywords.
Harvard calendars list hundreds of events across schools and departments. Reporters and students often:
- Miss relevant events because they don’t know what to search for.
- Rely on brittle keyword filters that catch too much or too little.
- Need to frequently change interests (e.g., tech → religion → student life).
This project solves that by providing topic-based, AI-powered alerts that are easy to update and require minimal manual effort.
Users define a topic they care about (e.g., technology, religious life, AI ethics). The system:
- Reads event titles and descriptions.
- Uses AI embeddings to understand semantic meaning.
- Alerts users about events that are relevant in substance, not just wording.
Example: A talk titled "Faith and the Law in Pluralistic Societies" would be surfaced for a topic like religious events, even if the word "religion" is never used.
Topics are designed to be easily changeable.
- Covering tech: Alerts for AI, startups, cybersecurity, data policy.
- Switching to religion: Alerts for chaplain talks, faith-based panels, moral philosophy.
When a relevant event is detected, the system sends an email containing:
- Event title
- Date and time
- Location
- Hosting organization
- Link to the calendar listing
This is a prototype, not a production system. Success is defined as:
- Accurately surfacing relevant events.
- Allowing easy topic changes.
- Delivering clean, useful email alerts.
Note: Continuous monitoring is a bonus feature, not a core requirement.