Query Graph Random Walk Ranking With Reinforcement Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional heuristic methods for processing user queries on exchange platforms fail to consider contextual or accumulated user information, user feedback, and do not optimize directional edges in computational graphs, leading to inaccurate and inefficient output predictions.
Innovation Solution
Implement reinforcement learning techniques to perform multiple random walks starting at both primary and secondary nodes in a computational graph, using parallel processing to boost scores of listing nodes based on contextual information and user interactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple random walks are performed in parallel starting at primary and secondary nodes, then the accuracy of determining listing nodes is improved, but the device complexity increases
Solution Approach 1:
The system segments the random walk process into multiple independent parallel executions, each starting from different nodes (primary and secondary). Each parallel thread performs random walks independently, and the results are aggregated to determine final listing nodes. This segmentation enables improved accuracy through multiple sampling paths while managing complexity through independent, modular execution units.
Solution Approach 2:
The system transitions from single-node random walks to multi-node parallel random walks by adding the dimension of parallel execution. Multiple threads execute simultaneously on different processing cores, exploring the computational graph from multiple starting points (primary and secondary nodes) at the same time. This dimensional expansion increases accuracy without proportionally increasing sequential complexity.
2Productivity
If ten times more random walk samplings are performed within a given latency, then the productivity increases, but the use of energy increases
Solution Approach 1:
The system performs random walk samplings in periodic batches across multiple parallel threads rather than continuously in a single sequence. Each thread performs a configured number of random walks, then results are aggregated. This periodic parallel execution achieves higher total sampling throughput while managing energy consumption through structured, interruptible execution cycles that can be optimized based on resource availability.
Solution Approach 2:
The system changes the parameter of parallelism degree by utilizing multiple processing cores simultaneously. By distributing random walk executions across available CPU cores, the system achieves ten times more samplings within the same latency window. The energy cost is distributed across multiple cores rather than concentrated in a single overloaded processor, allowing for more efficient energy utilization through parallel architecture.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for processing a user query, comprising receiving a user query corresponding to a primary node in the computational graph; processing the user query to determine one or more listing nodes in the computational graph, providing, to the user device, data associated with one or more listing nodes. The processing comprises: determining a first set of random walks starting at the primary node according to a policy, determining a second set of random walks starting at one or more secondary nodes according to the policy, determining a final score for each of a first set of listing nodes reached by the first set of random walks, and determining the one or more listing nodes from the first set of listing nodes based on the final scores.