TV Series Recommendation System using PageRank Algorithm

Project Overview

The goal of this project is to develop a recommendation system for TV series by leveraging the PageRank algorithm. The system identifies the most influential genres in a dataset of TV series and uses these insights to recommend series based on genre relevance.

---

Steps Involved

1. Loading the Dataset

We used a dataset containing the following fields:

• Series Title
• Release Year
• Runtime
• Genre
• Rating
• Cast
• Synopsis

Example row:

Title	Release Year	Runtime	Genre	Rating	Cast	Synopsis
Wednesday	2022–	45 min	Comedy, Crime, Fantasy	8.2	Jenna Ortega, Hunter Doohan	Follows Wednesday Addams' years as a student...

---

2. Creating the Graph Structure

• Nodes: Represent genres.
• Edges: Represent co-occurrences of genres in the same series.
• Construction:
  • Split the Genre field into a list of genres for each series.
  • Connect each genre to every other genre in the list, forming a graph of genre co-occurrences.

---

3. Implementing the PageRank Algorithm

• Initialization: Assign an equal initial PageRank score to each genre.
• Iterations:
  • Update each genre's score based on contributions from connected genres.
  • Use a damping factor d = 0.85 to account for random jumps between nodes.
• Convergence: Continue iterating until the score changes between iterations are below a threshold (e.g., 1e−6).

---

4. Integrating PageRank Scores into Recommendations

• Genre Importance: Use the PageRank scores to evaluate the importance of each genre.
• Recommendation Score:
  • For a given series, sum the PageRank scores of its genres.
  • Rank series based on their recommendation scores.
• Output: Return the top-ranked series as recommendations.

---

5. Interpreting the Recommendations

• The system identifies series with genres that have high PageRank scores, reflecting influential and well-connected genres.
• The recommendations aim to highlight series with prominent genres.

---

Benefits and Limitations

Benefits

• Genre-Based Insights: Highlights series with influential and commonly occurring genres.
• Simplicity: Focuses solely on genre co-occurrence, making it computationally lightweight.

Limitations

• Lack of Personalization: Does not consider user preferences or past interactions.
• Limited Metadata Usage: Only uses genre information; additional factors like ratings or detailed metadata are not included.

---

Usage

To use the recommendation system:

1. Load a dataset with TV series information (including genres).
2. Run the script to construct the graph and compute PageRank scores.
3. Input a series or genre to receive top recommendations.

---