Multi-level cross-view contrastive learning for enhanced item intention-aware recommender system

Key Points

• Item intention is better predicted through multi-level contrastive learning techniques, allowing for personalized recommendations.
• The model effectively uses contrastive learning to improve item intention awareness, resulting in a substantial 25% increase in recommendation accuracy.
• Analysis employs a novel feature extraction approach to derive user preferences while accounting for varying intents in product searches.
• This method highlights the need for improving personalization in recommendation systems, emphasizing user intent to boost engagement.