Content Recommendation Engine for Mobile Bounce Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mobile users often 'bounce back' quickly from third-party websites after visiting, leading to high bounce rates that are undesirable for publishers, as users typically return to the application or service that provided the link rather than exploring the third-party site further.
Innovation Solution
A system that provides content recommendations to users based on their interest profiles and site interactions, using a processor and memory to retrieve candidate items from various sources such as social, authorship, relevant, and popular sources, and ranks them for presentation as recommendations, thereby encouraging users to engage more with the site.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If content recommendations are provided to users on a site, then page views per visit increase and bounce rates decrease, but system complexity increases due to multiple data sources and processing requirements
Solution Approach 1:
The system segments the recommendation generation process into distinct modules: candidate item retrieval from multiple sources (social graph, authorship, relevance, popularity), scoring module that evaluates candidates against user interest profiles, and ranking module that orders results. This segmentation allows each component to be optimized independently while managing overall system complexity.
Solution Approach 2:
The patent introduces an intermediary recommendation engine that sits between the user and the content sources. This intermediary processes user interest profiles, queries multiple data sources, and presents synthesized recommendations, thereby coordinating complex interactions without requiring direct integration between all components.
2Productivity
If personalized content recommendations are generated using user interest profiles and multiple data sources, then user engagement increases, but data processing time and computational resources increase
Solution Approach 1:
The system performs preliminary actions by pre-computing and storing user interest profiles based on historical behavior, and by pre-organizing content from multiple sources into accessible formats. When generating recommendations, the system queries these pre-processed data structures rather than analyzing raw data from scratch, significantly reducing real-time processing requirements.
Solution Approach 2:
The recommendation system retrieves a larger set of candidate items from multiple sources than will ultimately be displayed, then applies scoring and filtering to select the most relevant subset. This approach allows the system to explore diverse content sources thoroughly while ensuring only the best matches are presented to users, balancing comprehensiveness with efficiency.
Data Source
AI summary
The disclosure includes a system and method for providing content recommendation to users on a site. The system includes a processor and a memory storing instructions that, when executed, cause the system to: receive data describing a page on a site visited by a first user; query one or more sources to retrieve one or more candidate items for the first user; determine one or more source types related to the one or more candidate items; determine one or more scores for the one or more candidate items based on an interest profile and the one or more source types; rank the one or more candidate items; determine that an invitation trigger event causing a presentation of an invitation state occurs; determine that a recommendation trigger event causing a presentation of a recommendation state occurs during the invitation state; and provide a recommendation state to the first user.