ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT

TL;DR

ColBERT introduces a late interaction architecture for BERT-based passage search that significantly reduces computational costs while maintaining high effectiveness, enabling fast and scalable information retrieval.

Contribution

The paper presents ColBERT, a novel BERT-based ranking model with a late interaction mechanism that allows pre-computation of document representations for efficient retrieval.

Findings

ColBERT achieves competitive effectiveness with existing BERT models.

It is two orders of magnitude faster than traditional BERT re-ranking models.

Requires four orders-of-magnitude fewer FLOPs per query.

Abstract

Recent progress in Natural Language Understanding (NLU) is driving fast-paced advances in Information Retrieval (IR), largely owed to fine-tuning deep language models (LMs) for document ranking. While remarkably effective, the ranking models based on these LMs increase computational cost by orders of magnitude over prior approaches, particularly as they must feed each query-document pair through a massive neural network to compute a single relevance score. To tackle this, we present ColBERT, a novel ranking model that adapts deep LMs (in particular, BERT) for efficient retrieval. ColBERT introduces a late interaction architecture that independently encodes the query and the document using BERT and then employs a cheap yet powerful interaction step that models their fine-grained similarity. By delaying and yet retaining this fine-granular interaction, ColBERT can leverage the expressiveness of deep LMs while simultaneously gaining the ability to pre-compute document representations offline, considerably speeding up query processing. Beyond reducing the cost of re-ranking the documents retrieved by a traditional model, ColBERT's pruning-friendly interaction mechanism enables leveraging vector-similarity indexes for end-to-end retrieval directly from a large document collection. We extensively evaluate ColBERT using two recent passage search datasets. Results show that ColBERT's effectiveness is competitive with existing BERT-based models (and outperforms every non-BERT baseline), while executing two orders-of-magnitude faster and requiring four orders-of-magnitude fewer FLOPs per query.