Bridge Item Recommendation Strategy for Video Content
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current recommendation systems in video delivery services often struggle to recommend content that users are unlikely to engage with in the short term, leading to missed opportunities for long-term user retention and engagement.
Innovation Solution
The implementation of a bridge item recommendation strategy, where a bridge item is recommended instead of the target item, with the goal of increasing the likelihood that the user will subsequently engage with the target item.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the recommendation system recommends content based on user's watch history and characteristics, then the user is likely to engage with the recommended content in the short term, but the recommendations are limited to a similar scope and do not achieve long-term user retention
Solution Approach 1:
The patent introduces a bridge item as an intermediary between the user's current interests (watch history) and the target item (desired long-term retention content). The bridge item recommendation engine selects items that serve as stepping stones, gradually guiding users from their current content preferences toward target content they might not directly choose but would be valuable for long-term retention. This intermediary approach resolves the contradiction by maintaining short-term engagement through familiar content while progressively expanding content scope diversity.
2Measurement precision
If the recommendation system focuses on short-term user interests, then the recommendations are highly relevant to current user preferences, but the system misses opportunities for long-term user retention and engagement
Solution Approach 1:
The patent implements preliminary action by proactively recommending bridge items that prepare users for future target content consumption. Instead of waiting for users to naturally evolve their interests, the system pre-positions bridge items in the recommendation stream that will gradually guide users toward target content. This preliminary guidance maintains high recommendation relevance while extending user retention period by strategically planning the user journey toward valuable long-term content.
3Ease of operation
If the recommendation system recommends only content similar to user's watch history, then the user is likely to watch the recommended content, but the system cannot introduce users to new content they may not have chosen otherwise
Solution Approach 1:
The patent segments the recommendation pathway into multiple stages: bridge items that are similar to user's watch history (easy to accept) and target items that represent desired long-term content (harder to reach directly). The bridge item recommendation engine creates a segmented transition path, breaking down the gap between user current preferences and target content into manageable steps. This segmentation maintains ease of operation at each step while achieving content discovery capability through the cumulative effect of the recommendation sequence.
Data Source
AI summary
In some embodiments, a method identifies a target item. A first score for a bridge item is generated using a first input where the first input is based on a characteristic of an entity. A second score for the target item is generated using a second input. The second input is based on the characteristic of the entity and the bridge item. The method determines whether to output the bridge item to the entity based on the first score and the second score.