A Semantic Search Pipeline for Causality-driven Adhoc Information Retrieval

2025-03-02

Fire

AI Summary

This paper addresses the causal-aware ad-hoc retrieval (CAIR) task, where conventional information retrieval (IR) methods retrieve only topically relevant documents and fail to identify causal relationships between events. To tackle this, we propose the first unsupervised semantic search framework that disentangles causal intent from topical intent to enable causality-directed retrieval. Our approach synergistically integrates BERT-based semantic indexing with BM25 lexical indexing, and introduces three novel components: (i) event causal pattern expansion, (ii) counterfactual query reconstruction, and (iii) causality-aware re-ranking. Evaluated on the CAIR-2021 benchmark, our method achieves state-of-the-art performance, outperforming traditional IR baselines and pure semantic embedding approaches by over 12% in NDCG@10. This work provides the first empirical validation of the effectiveness and superiority of unsupervised paradigms for fine-grained causal retrieval.

Technology Category

• Reasoning under Uncertainty: Causality
• Machine Learning: Causal Learning
• Data Mining & Knowledge Management: Intelligent Query Processing

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs

Abstract

We present a unsupervised semantic search pipeline for the Causality-driven Adhoc Information Retrieval (CAIR-2021) shared task. The CAIR shared task expands traditional information retrieval to support the retrieval of documents containing the likely causes of a query event. A successful system must be able to distinguish between topical documents and documents containing causal descriptions of events that are causally related to the query event. Our approach involves aggregating results from multiple query strategies over a semantic and lexical index. The proposed approach leads the CAIR-2021 leaderboard and outperformed both traditional IR and pure semantic embedding-based approaches.

Problem

• Develops a semantic search pipeline for causality-driven information retrieval.
• Distinguishes topical documents from causal event descriptions.
• Outperforms traditional and semantic embedding-based retrieval methods.

Innovation

• Unsupervised semantic search pipeline for causality-driven retrieval
• Aggregates results from multiple query strategies
• Combines semantic and lexical indexing for improved accuracy