Ted Talks Recommendation System with Machine Learning

This project is a Flask-based web application that uses Natural Language Processing (NLP) and machine learning to recommend TEDx talks related to a user's input. The system analyzes the meaning of the text provided and finds the most relevant talks, displaying the main speaker and a short description for each recommendation.

How it works

The app first cleans and normalizes the text, removing punctuation and irrelevant words (like "the" or "and"), to focus on meaningful terms. Then, it uses TF-IDF (Term Frequency – Inverse Document Frequency) to transform the text into numerical values, assigning higher importance to unique or significant words. For example, in "technological innovation for social change," words like "innovation" and "technological" are weighted more heavily than common terms.

To measure relevance, the system combines Cosine Similarity (to assess how closely two texts "point" in the same direction in a mathematical space) and Pearson Correlation (as a statistical check for consistency). Finally, it ranks and returns the most relevant TEDx talks.

Example use case

If a user writes:

"I want to learn about leadership and personal motivation."

The system might recommend talks on:

• Leadership strategies for teams.

• Inspiring stories of overcoming challenges.

• Practical tips for personal growth.

Applications beyond TEDx

This recommendation approach is highly adaptable and can be applied across multiple industries:

• Education: Suggesting personalized courses or resources based on a student's needs.

• Human Resources: Matching employees with relevant training materials or internal guides.

• E-commerce & Marketing: Recommending products based on a customer's natural language descriptions.

• Healthcare: Suggesting validated articles or advice based on symptoms or health questions.

• Customer Support: Powering bots that instantly find relevant knowledge base articles for common issues.

Technologies used

• Backend: Flask (Python)

• Text processing: NLTK

• Similarity models: Scikit-learn (TF-IDF, Cosine Similarity, Pearson Correlation)

• Frontend: HTML + JavaScript with AJAX for dynamic responses

Business value

This type of system helps organizations save time, personalize user experiences, and scale efficiently. Whether it's connecting students with learning materials, helping employees access resources, or improving product recommendations, intelligent text-based recommendations deliver measurable impact.