TechPrufer
Get started
Qdrant
Qdrant logoQdrant · Feature trail

Relevance Score Boosting and Decay Functions

2 posts · Apr 2025 – Feb 2026

What this trail covers

Qdrant enhances its relevance scoring capabilities by detailing the implementation and usage of decay functions (Linear, Exponential, Gaussian) within its score boosting functionality. These functions allow numeric payload fields or similarity scores to influence search relevance by transforming values into a 0.0-1.0 scale based on a target value and decay parameters (scale, midpoint). The post provides examples for use cases like time-based relevance, distance-based relevance, and price sensitivity. The Qdrant 1.14 release introduces a Score-Boosting Reranker that allows blending vector similarity with custom rules and context, enabling users to influence ranking using metadata like timestamps, location info, and numeric attributes. This allows for prioritizing content, recency, or geographical proximity in search results.

All Qdrant trails
  1. Qdrant 1.17 - Relevance Feedback & Search Latency Improvements - QdrantIntroduces Relevance Feedback Query, a new method to refine search results using lightweight feedback on result relevance. Implements search latency improvements including an update queue to manage write load and delayed fan-outs to reduce tail latency. Enhances operational observability with a cluster-wide telemetry API and segment optimization monitoring. Redesigns the Web UI point search interface.
  2. Untangling Relevance Score Boosting and Decay Functions - QdrantThis post details the implementation and usage of Linear, Exponential, and Gaussian decay functions within Qdrant's score boosting functionality. It explains the purpose of these functions in transforming numeric properties into relevance scores (1.0 to 0.0) and influencing final relevance. The post clarifies the parameters: 'x' (input variable), 'target' (ideal value for 1.0 relevance), 'scale' (controls decay rate with 'x'), and 'midpoint' (relevance value at 'target ± scale'). It provides examples for use cases like video length, promo code freshness, and price sensitivity, and explicitly states that 'midpoint' and 'scale' are static and cannot be dynamically set during query time, warning against dynamic normalization of scores based on query subsets.
  3. Qdrant 1.14 - Reranking Support & Extensive Resource Optimizations - QdrantIntroduces a Score-Boosting Reranker that allows blending vector similarity with custom rules and context. This feature enables users to influence ranking using metadata like timestamps, location info, and numeric attributes. Examples include prioritizing website content based on tags (title, paragraph), reranking recent results using datetime fields, and factoring in geographical proximity using decay functions. The post also details incremental HNSW indexing for faster indexing of new points, improved parallel processing for batch queries, and optimizations for segment optimization and memory usage in immutable segments.