Improving Ad matching via Cluster-Adaptive Keyword Expansion and Relevance tuning

2025-05-24

AI Summary

In search advertising, token-based semantic expansion often improves keyword coverage at the expense of relevance. This paper proposes a lightweight, document-side semantic expansion framework: user queries remain unchanged, while a pretrained siamese language model generates dense representations for ad documents; a local-density-aware clustering method adaptively determines similarity thresholds to control expansion granularity. An incremental decision tree ensemble model further recalibrates relevance ranking. Our approach pioneers the synergistic integration of clustering-adaptive thresholding and document-side incremental semantic adaptation—achieving high recall and high relevance without increasing query-side computational complexity. Experiments demonstrate significant improvements in ad relevance and click-through rate (CTR), with support for low-latency, scalable deployment. The framework effectively handles dynamic query distributions and evolving ad inventories.

Technology Category

• Search and Optimization: Distributed Search
• Machine Learning: Learning Preferences or Rankings
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• Search and Retrieval-Augmented AI: Ad search and search for Web retail
• User Modeling, Personalization and Recommendation: User modeling for targeted and personalized online advertising
• Semantics and Knowledge: Applications of semantic technologies for improving search, browsing, recommendation, personalization

Abstract

In search advertising, keyword matching connects user queries with relevant ads. While token-based matching increases ad coverage, it can reduce relevance due to overly permissive semantic expansion. This work extends keyword reach through document-side semantic keyword expansion, using a language model to broaden token-level matching without altering queries. We propose a solution using a pre-trained siamese model to generate dense vector representations of ad keywords and identify semantically related variants through nearest neighbor search. To maintain precision, we introduce a cluster-based thresholding mechanism that adjusts similarity cutoffs based on local semantic density. Each expanded keyword maps to a group of seller-listed items, which may only partially align with the original intent. To ensure relevance, we enhance the downstream relevance model by adapting it to the expanded keyword space using an incremental learning strategy with a lightweight decision tree ensemble. This system improves both relevance and click-through rate (CTR), offering a scalable, low-latency solution adaptable to evolving query behavior and advertising inventory.

Problem

• Expanding ad keyword reach without losing relevance
• Adjusting similarity thresholds based on semantic density
• Enhancing relevance models for expanded keyword space

Innovation

• Document-side semantic keyword expansion using language model
• Cluster-based thresholding for similarity adjustment
• Incremental learning with lightweight decision tree ensemble