Social Graph Traversal for Ad Targeting Expansion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Social networking systems limit the reach of advertisements by relying solely on user characteristics for targeting, failing to leverage other information logged in the system to expand the target audience.
Innovation Solution
A social networking system allows advertisers to seed a search within the social graph to identify targeting criteria, using keywords or other objects associated with the ad, and iteratively selects candidate objects based on similarity to expand the ad's reach by traversing the graph and presenting additional candidate objects for selection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If targeting criteria are limited to user characteristics only, then the targeting process is simple and direct, but the ad reach is restricted to a smaller group of users
Solution Approach 1:
The targeting process is segmented into multiple stages: initial user characteristic filtering, seed object identification, candidate object generation through graph traversal, and iterative refinement. This segmentation allows the system to handle large-scale data processing in manageable steps while expanding ad reach beyond simple demographic targeting.
Solution Approach 2:
The system transitions from one-dimensional targeting (user characteristics only) to multi-dimensional targeting by incorporating social graph relationships, object similarities, and interaction patterns. This dimensional expansion enables the system to identify users based on diverse criteria including social connections, content interactions, and contextual relationships, thereby significantly increasing ad reach.
2Measurement precision
If the system traverses the social graph to identify candidate objects, then the ad targeting becomes more precise and comprehensive, but the processing time and computational resources increase
Solution Approach 1:
The system performs preliminary actions by pre-processing and indexing social graph data, user profiles, and object metadata before ad targeting is needed. This includes pre-computing similarity metrics, organizing graph traversals, and preparing candidate object pools, which significantly reduces processing time during actual ad delivery while maintaining high targeting precision.
Solution Approach 2:
The system employs partial graph traversal and selective candidate generation rather than exhaustive processing of the entire social graph. By using thresholds for similarity scores, limiting traversal depth, and prioritizing high-relevance connections, the system achieves sufficient targeting precision without the computational burden of complete graph analysis, thus reducing processing time.
3Adaptability or versatility
If the system presents multiple candidate objects for advertiser selection, then the ad targeting becomes more flexible and adaptable, but the user interface complexity increases
Solution Approach 1:
The interface incorporates feedback mechanisms where advertisers can review, refine, and adjust candidate object selections in real-time. The system provides feedback on candidate relevance, allows iterative selection modifications, and adapts to advertiser preferences, maintaining flexibility while simplifying the selection process through intelligent guidance and automated filtering.
Solution Approach 2:
The system introduces intermediary layers between the advertiser and the complex social graph data. This includes automated candidate generation, relevance ranking, and filtering mechanisms that translate complex graph relationships into user-friendly presentations. The intermediary layer abstracts away the complexity of graph traversal and similarity calculations, presenting advertisers with curated, actionable results.
Data Source
AI summary
A social networking system extends the reach of an ad by allowing an advertiser to seed a search for suggesting targeting criteria for an ad. Using the seeds, the social networking system identifies a set of seed objects from a social graph. The social networking system traverses the social graph to compute the similarity between the seed objects and candidate objects related to the seed objects. Based on the computed similarity, the social networking system suggests a set of target objects to the advertiser. The social networking system receives a selection of one or more target objects from the advertiser. From the selected target objects the social networking system determines one or more targeting criteria for the ad, so users with an edge in the social graph to any of the selected target objects are eligible to receive the ad.