Social Context Mapping for External Ad Targeting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Social networking systems struggle to effectively target advertisements linking to external content since they lack social context and affinities for content outside their system, preventing accurate presentation to interested users.
Innovation Solution
A social networking system generates an index associating internal objects with external links, parsing these links into components like domain, path, and query string, and uses this information to identify candidate objects and apply rules for selecting the most relevant object to associate with the advertisement, thereby providing social context for external content.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the social networking system presents advertisements linking to external content, then the system can expand its advertising capabilities beyond internal content, but the system cannot determine social context or affinities for the external content, preventing effective targeting
Solution Approach 1:
The patent introduces an intermediary mapping mechanism that connects external content to internal social graph objects. When an external advertisement is encountered, the system parses the URL and searches for corresponding internal objects (pages, applications, etc.) that serve as intermediaries. This mapping allows the system to leverage existing social context and affinity data about internal objects to target external advertisements, effectively bridging the information gap without requiring direct social context for external content.
2Loss of information
If the system creates a mapping between internal objects and external links, then social context can be associated with external content, but the system complexity increases due to parsing and indexing requirements
Solution Approach 1:
The patent segments the URL into distinct components (domain, path, query string) and creates mappings at different levels of granularity. This segmentation allows the system to build indexes organized by URL components, enabling efficient partial matching. For example, if an exact URL match fails, the system can still match on domain or path components alone, reducing the complexity of requiring perfect matches while maintaining mapping effectiveness.
Solution Approach 2:
The system performs preliminary parsing of URLs and pre-computation of mappings between internal objects and external links during off-line processing. By preparing these mappings in advance and storing them in indexes, the system avoids complex real-time parsing and matching operations when serving advertisements. This preliminary action reduces online computational complexity while maintaining comprehensive mapping coverage.
3Measurement precision
If the system parses links into components and creates detailed indexes, then accurate matching can be achieved, but the processing time and computational resources increase
Solution Approach 1:
The patent implements a dynamic matching strategy that adapts the level of parsing detail based on the situation. The system first attempts quick matching using coarse-grained URL components (domain only), and only if that fails does it proceed to more detailed parsing (path, query string). This dynamic approach ensures high matching accuracy when needed while minimizing processing time for the majority of cases where simple domain matching suffices, effectively balancing precision and speed.
Data Source
AI summary
A social networking system associates one or more objects in the social networking system with a link to content external to the social networking system. This allows information about the object to be used as a proxy for information about the external content. An index associating objects in the social networking system with links to content external to the social networking system is maintained. A received link to external content are parsed into components, such as a domain name, a path, and a query string, and the components are compared to the index to identify an object associated with one or more of the comments of the received link. One or more rules may be applied to increase the likelihood of the identified object having similar subject matter as the external content.