Incorrect Definitions and Confusion Between User-User and Item-Item Collaborative Filtering

[jash0803]

Here's exactly what's wrong:

In Paragraph	Reality (Correct)
"User-Item Filtering: Also called item-item collaborative filter"	Wrong. User-item filtering is the general matrix of interactions, not item-item filtering.
"Item-Item Filtering: Also called user-user collaborative filter"	Wrong. Item-item filtering is about finding similar items, not similar users.

---

Short Summary of Issues:

1. Terminology confusion: "User-Item Filtering" ≠ "Item-Item Collaborative Filtering"

2. Concept reversal: Item-Item is mistakenly described as User-User and vice versa.

3. Misleading for readers: It would confuse people learning recommender systems, leading to incorrect implementations.

---

There is a critical mistake in the documentation regarding collaborative filtering types:

• The paragraph incorrectly states that:

  • User-Item Filtering is also called Item-Item Collaborative Filtering, and

  • Item-Item Filtering is also called User-User Collaborative Filtering.

This is factually incorrect. Here's the correct interpretation:

• User-Item Filtering: Refers to the user-item interaction matrix itself (raw ratings or interactions), not a filtering technique.

• User-User Collaborative Filtering: Finds similar users based on their item preferences and recommends items that similar users liked.

• Item-Item Collaborative Filtering: Finds similar items based on user ratings and recommends items similar to the ones a user liked.

Suggested Fix: Correct the definitions to match standard machine learning and recommender system terminology. Swap the incorrectly labeled sections and provide accurate descriptions.

Impact: This confusion can cause developers and learners to misunderstand collaborative filtering techniques, leading to incorrect implementations and degraded recommendation quality.