Artificial Intelligence in Commerce: Recommenders & Dynamic Pricing

Explanation

AI recommendation systems generate up to 35% of revenue on leading platforms like Amazon. Collaborative filtering constructs a sparse User-Item interaction matrix R. In user-based filtering, similarity between users u and v is computed via Cosine Similarity or Pearson Correlation. The cold-start problem is mitigated through hybrid systems that combine collaborative filtering with content-based metadata and popular trending fallbacks. Dynamic pricing algorithms optimize revenue by predicting price elasticity of demand in real time.

Key points

• Cold-Start Problem: Challenge of generating recommendations for brand-new users or newly listed products with zero interaction history.
• Dynamic Pricing Algorithms: Machine learning models adjusting prices in real time based on demand elasticity, competitor rates, inventory levels, and customer browsing history.
• Visual Search: Convolutional neural networks allowing shoppers to upload photos to find identical or aesthetically similar clothing/furniture items.
• Conversational AI Chatbots: Large language model assistants providing personalized product discovery, sizing recommendations, and order status updates.

Example

Code examples

import numpy as np

# Cosine Similarity for Collaborative Filtering
def cosine_similarity(u, v):
    dot_product = np.dot(u, v)
    norm_u = np.linalg.norm(u)
    norm_v = np.linalg.norm(v)
    if norm_u == 0 or norm_v == 0:
        return 0.0
    return dot_product / (norm_u * norm_v)

# User ratings vectors across 4 products
user_A = np.array([5, 3, 0, 1])
user_B = np.array([4, 3, 0, 2])
print("Similarity between User A and User B:", cosine_similarity(user_A, user_B))

Exam questions

• [10 Marks] Explain Collaborative Filtering vs Content-Based Filtering in recommendation systems. How is the Cold-Start problem solved?
• [6 Marks] Discuss how Dynamic Pricing algorithms operate in digital commerce. What are the economic and ethical considerations?