Latency-Efficient Machine-Trained Ranking Model for Resource Retrieval
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing enterprise resource retrieval systems rely on manual, ad hoc rules that are unreliable, not scalable, and resource-inefficient, making it difficult to accurately and efficiently compare the relevance of resources across different classes.
Innovation Solution
A system using a prefix tree data structure and a machine-trained ranking model, such as a gradient boosted decision tree, to retrieve and rank resource descriptors based on user queries, providing a scalable and efficient way to compare resource relevance across heterogeneous resource types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manually-derived ad hoc rules are used for each resource type, then resource retrieval can be performed, but the system becomes unreliable, not scalable, and resource inefficient
Solution Approach 1:
The patent applies universality by replacing multiple ad hoc rule sets with a single machine-trained ranking model that handles all resource types (files, people, bookmarks, applications) uniformly. The model takes query and resource descriptor inputs and produces ranked results across heterogeneous resource classes, eliminating the need for separate rule maintenance for each resource type while improving retrieval efficiency and scalability.
2Measurement precision
If a comprehensive ranking model is used to accurately compare resources across different classes, then retrieval accuracy improves, but computational latency increases
Solution Approach 1:
The patent applies parameter changes by optimizing the machine-trained ranking model's computational parameters. The model uses gradient boosted decision trees with carefully selected hyperparameters (number of trees, tree depth, learning rate) that balance accuracy and latency. Feature engineering also transforms resource descriptors into optimized numerical representations that accelerate computation while preserving relevance information, enabling accurate cross-resource-type ranking with acceptable latency.
Data Source
AI summary
A technique suggests resources within a particular environment. The technique operates by receiving a query via a search interface presentation, and using a prefix tree data structure to retrieve candidate resource descriptors that match the query. The candidate resource descriptors are data items that describe different types of candidate resources. The technique further includes: generating feature information that describes the candidate resource descriptors; using a machine-trained ranking model to generate scores associated with the candidate resource descriptors, based on the feature information; sorting the candidate resource descriptors based on the scores, to provide sorted resource descriptors; and generating output information that represents a selected set of resource descriptors drawn from the sorted resource descriptors, for presentation via the search interface presentation. In some implementations, the machine-trained model is a boosted decision tree model optimized to achieve a desired accuracy and latency.