Synthesizing Trajectory Queries from Examples

2026-02-16

AI Summary

This work addresses the challenge data scientists face in efficiently authoring and tuning fuzzy query programs with real-valued parameters over machine learning outputs such as video object trajectories. To overcome this, the authors propose Quivr, a framework that automatically synthesizes trajectory query programs from user-provided positive and negative examples. The key innovation lies in the integration of parameter space pruning and quantized semantic representations, which together significantly enhance both the efficiency and accuracy of synthesizing real-valued parameters. Evaluated on a benchmark of 17 tasks, Quivr successfully generates high-precision queries with substantially reduced synthesis time compared to baseline methods, demonstrating the effectiveness and scalability of the approach.

Technology Category

• Search and Optimization: Learning to Search
• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Image and Video Retrieval

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Data scientists often need to write programs to process predictions of machine learning models, such as object detections and trajectories in video data. However, writing such queries can be challenging due to the fuzzy nature of real-world data; in particular, they often include real-valued parameters that must be tuned by hand. We propose a novel framework called Quivr that synthesizes trajectory queries matching a given set of examples. To efficiently synthesize parameters, we introduce a novel technique for pruning the parameter space and a novel quantitative semantics that makes this more efficient. We evaluate Quivr on a benchmark of 17 tasks, including several from prior work, and show both that it can synthesize accurate queries for each task and that our optimizations substantially reduce synthesis time.

Problem

• trajectory queries
• program synthesis
• real-valued parameters
• fuzzy data
• query synthesis

Innovation

• trajectory query synthesis
• program synthesis
• parameter space pruning
• quantitative semantics
• example-based programming