Personal Knowledge Graph for News Feed Deduplication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing content filtering methods fail to effectively deduplicate and prioritize news feed content based on user-specific knowledge, leading to redundant information being presented to users.
Innovation Solution
A computer-implemented method that creates a personal knowledge graph for each user, determining story arcs and novelty scores of incoming content based on user interaction and consumption history, to filter and prioritize content accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If traditional deduplication methods are used based on character string matching, then redundant content can be suppressed, but user-specific knowledge and personalization are not considered
Solution Approach 1:
The system performs preliminary actions by building a user knowledge graph before content filtering occurs. This knowledge graph captures the user's existing knowledge, interests, and story arcs in advance, enabling personalized deduplication when new content arrives. The pre-built knowledge structure allows the system to quickly determine whether new content is redundant based on what the user already knows, rather than using generic string matching.
Solution Approach 2:
The user knowledge graph serves as an intermediary between incoming content and the filtering decision. Instead of directly comparing new content against all previous content, the system uses the knowledge graph as a mediator that represents the user's knowledge state. This intermediary structure enables personalized deduplication by translating user-specific knowledge into filterable representations.
2Measurement precision
If a comprehensive content analysis is performed to determine user knowledge, then personalization accuracy is improved, but computational complexity increases
Solution Approach 1:
The system segments the user's knowledge into discrete story arcs, each representing a specific topic or event. Instead of maintaining a monolithic knowledge base, the knowledge graph is divided into multiple story arc components, each with its own entities and relationships. This segmentation allows the system to analyze and filter content related to specific story arcs independently, reducing overall computational complexity while maintaining precision.
Solution Approach 2:
The system changes parameters by representing knowledge in a structured graph format with specific properties (entities, relationships, story arcs) rather than unstructured text. This parameterized representation enables efficient querying and comparison operations. The knowledge graph uses standardized parameters such as story arc identifiers, entity types, and relationship categories that facilitate fast matching and deduplication decisions.
3Reliability
If story arc-based deduplication is implemented, then content relevance is improved, but system complexity increases
Solution Approach 1:
The system performs preliminary story arc extraction and knowledge graph construction when users interact with content, before filtering new content arrives. By pre-processing and organizing knowledge into story arc structures in advance, the system reduces the complexity of real-time filtering decisions. When new content arrives, the system only needs to check against the pre-built story arc representations, which simplifies the filtering process.
Solution Approach 2:
The system creates simplified copies of user knowledge in the form of story arc representations. Instead of storing and processing all original user interactions, the system extracts essential information and creates compact story arc models that capture the essence of user knowledge. These copied representations are much easier to process and compare against new content, reducing system complexity while maintaining relevance.
Data Source
AI summary
In an approach to personalizing a news feed, a computing device monitors a user accessing content. The computing device determines a personal knowledge graph for the user based on the accessed content. Responsive to receiving a new piece of content that the user has not accessed, the computing device determines a novelty score for the new piece of content based on the personal knowledge graph. The computing device filters the new piece of content based on the novelty score.