Content-based filtering

lastmod 2024-04-24created 2024-04-247 edits

Content-based Filtering uses the domain of the items themselves: it analyzes the attributes of items a user is interested in and recommends new items.

It differs from Collaborative Filtering (Item-based) in that information about other users is not used.

About

For example, it is often used to recommend clothes that look similar to clothes I bought, or to recommend other news similar or related to a news article.

Content-based filtering uses Feature Extraction methods such as TF-IDF (Term Frequency - Inverse Document Frequency) or Word2Vec to extract item features well. Choosing the Similarity measure for comparing items using the extracted features is also important.

• Pros: Can recommend from item information alone, without information about user preferences
• Cons: Requires building data that describes the items (item's metadata).

Categories

• TF-IDF
• Word2Vec

Content-based filtering vs Collaborative filtering

See Recommender system#Content-based filtering vs Collaborative filtering.

See also

• Recommender system (Korean: 추천 시스템)
• Content-based filtering (Korean: 컨텐츠 기반 필터링)
• Collaborative filtering (Korean: 협업 필터링)
• Hybrid recommender systems
• Learning to rank (machine-learned ranking; MLR; Korean: 순위 학습)
• Similarity (Korean: 유사도)

Favorite site

• Content-based Filtering | Machine Learning | Google for Developers
• 02. Content-based recommender systems - similarity functions and evaluation functions
• 02. Content-based recommender systems - a recommender system using TF-IDF
• 02. Content-based recommender systems - a recommender system using Word2Vec