Concept Graph User Intent Extraction for Social Media
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for understanding and creating temporally evolving user profiles in large-scale social media platforms are limited by their inability to effectively capture and represent individual and collective user behavior from diverse, unstructured data, leading to inaccurate and outdated user intent profiles.
Innovation Solution
A system and methodology using a global concept graph with nodes representing concepts and edges as relationships, allowing for the creation of unified, granular intent profiles by aggregating user activities and mapping them to weighted subgraphs, enabling accurate and detailed user intent capture and comparison across time and locations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If detailed structured databases are created based on explicitly stated user attributes, then user profile information is easy to collect and store, but the expressive capabilities are limited and the information becomes outdated and noisy
Solution Approach 1:
The patent introduces a concept graph as an intermediary structure that bridges explicit user attributes and implicit user intents. The concept graph contains nodes representing concepts (entities, events, activities) and edges representing relationships between them. User actions are mapped to concepts in the graph, allowing the system to infer detailed user intents and preferences that go beyond explicitly stated attributes, thereby resolving the contradiction between ease of data collection and accuracy of user intent characterization
Solution Approach 2:
The patent replaces the traditional mechanical approach of manually categorizing user actions into predefined taxonomies with an automated semantic mapping mechanism. Instead of requiring manual classification of billions of user actions, the system automatically maps user actions to concepts in the concept graph based on semantic relationships, enabling accurate and scalable user intent extraction without the limitations of manual taxonomy-based approaches
2Device complexity
If taxonomies with predefined categories are used to classify user actions, then classification structure is established, but the categories are limited and manual classification is highly error-prone
Solution Approach 1:
The patent creates a universal concept graph that can accommodate diverse user actions and content types without requiring separate taxonomies for each domain. The concept graph serves multiple functions: it structures user profiles, classifies user actions, infers user intents, and enables personalized recommendations. This multi-functional approach eliminates the need for multiple limited taxonomies and their associated manual classification processes, thereby improving classification accuracy while maintaining organizational structure
Solution Approach 2:
The system enables self-service classification by automatically mapping user actions to concepts in the concept graph without requiring manual intervention. The automated mapping process uses semantic relationships between user actions and graph concepts to perform classification, eliminating the errors and limitations of manual taxonomy-based classification while maintaining the benefits of structured organization
3Device complexity
If manual training and keyword-based classification are used for taxonomy categorization, then classification framework is established, but the process is computationally intensive and accuracy is limited by training quality
Solution Approach 1:
The patent extracts the essential semantic meaning of user actions by mapping them to concepts in the concept graph, rather than requiring full manual training or extensive keyword matching. This extraction approach captures the core intent of user actions efficiently, eliminating the computational burden of training complex classification models on billions of actions while maintaining accurate classification through semantic relationships in the graph
Data Source
AI summary
Methods and systems for extracting intents and intent profiles of users, as inferred from the different activities they execute and data they share on social media sites, and then (i) monetization of such intents via targeted advertisements, and (ii) enhancement of user experience via organization of their contact lists and conversations and posts based on their content and conceptual context.