Temporal Knowledge Graph for POI Recommendation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing point-of-interest recommendation methods fail to fully utilize multimodal information and effectively divide behavior pattern information, leading to inadequate learning of personalized user preferences.
Innovation Solution
A point-of-interest recommendation method based on a temporal knowledge graph is developed, constructing dynamic and static knowledge graphs to learn user behavior patterns and fuse multimodal information using a heterogeneous mutual attention mechanism and aspect-level sentiment analysis, enabling accurate prediction of next visited locations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple factors (trajectory information, geographic location, social network, user reviews) are utilized for POI recommendation, then the prediction capability is improved, but the device complexity increases due to the need to process and fuse multiple modalities of information
Solution Approach 1:
The patent segments the complex information fusion task into distinct modules: a graph construction module that builds knowledge graphs from multiple data sources, a graph neural network module that processes the structured relationships, and a recommendation module that generates predictions. This segmentation allows each module to handle specific aspects of multimodal information independently, reducing overall system complexity while maintaining comprehensive prediction capability.
Solution Approach 2:
The patent introduces knowledge graphs as an intermediary structure to bridge multiple information modalities. The knowledge graph serves as a unified representation that integrates trajectory information, geographic location, social network, and user reviews into a structured format with explicit relationships, enabling the graph neural network to process diverse data types through a common interface without requiring complex fusion logic.
2Measurement precision
If behavior pattern information in historical trajectory is not effectively divided, then the model simplicity is maintained, but the learning of personalized user preferences becomes insufficient
Solution Approach 1:
The patent segments historical trajectory data into meaningful behavior patterns by constructing knowledge graphs that identify and separate different types of user behaviors (e.g., sequential visits, revisits, diverse exploration). The graph neural network then processes these segmented patterns independently, allowing the model to learn personalized preferences from distinct behavior types without requiring overly complex processing mechanisms.
Data Source
AI summary
The disclosure discloses a point-of-interest recommendation method based on a temporal knowledge graph. Based on the historical behavior trajectory of the user and multimodal information, the disclosure constructs a dynamic temporal knowledge graph and a static group knowledge graph, which are used to learn interest preferences of the user that change over time and stable features that do not change over time, respectively. At the same time, the disclosure uses deep learning methods to build a point-of-interest recommendation model, which can extract a point-of-interest fusion feature representation and a user fusion feature representation from the two knowledge graphs in combination with a user review sentiment embedding sequence to accurately predict a point of interest most likely to be visited by the target user at a next moment. The disclosure has the characteristics of high precision and strong scalability, and can provide support for personalized user behavior trajectory prediction.