A recommendation system using a machine learning algorithm gives users the preferred product using a content-based algorithm to satisfy customer preferences, which is well served, and from the user's consumption record, it keeps giving out the same product that has already been satisfied, which then reduces user experience. Will this increase the e-commerce recommendation, or do we need to focus more on the knowledge base approach?
Emre Kuru
The problem you're referring to is a lack of serendipity in recommender systems. As you said, it's not enough to simply elicit the user's preferences with items they love the most, you also need to recommend items they haven't seen yet. To add serendipity to your system, you can:
- Utilize knowledge-based recommendations
- Incorporate simple popularity-based recommendations
- Employ a hybrid approach combining collaborative and content-based recommendations
0 votes
0 thanks
- Badges
- Science topic
- Similar topics
- Social Science
- Media
- Journalism
More Simon Eisabor's questions See All
Similar questions and discussions