Efficiency and Effectiveness of SPLADE Models on Billion-Scale Web Document Title
This work addresses efficiency challenges for deploying sparse retrieval models in large-scale search engines, offering incremental improvements.
This paper compared BM25, SPLADE, and Expanded-SPLADE models for large-scale web document retrieval, finding that SPLADE and Expanded-SPLADE outperformed BM25, especially on complex queries, but with higher computational costs; pruning strategies were introduced to improve efficiency, with Expanded-SPLADE achieving the best balance.
This paper presents a comprehensive comparison of BM25, SPLADE, and Expanded-SPLADE models in the context of large-scale web document retrieval. We evaluate the effectiveness and efficiency of these models on datasets spanning from tens of millions to billions of web document titles. SPLADE and Expanded-SPLADE, which utilize sparse lexical representations, demonstrate superior retrieval performance compared to BM25, especially for complex queries. However, these models incur higher computational costs. We introduce pruning strategies, including document-centric pruning and top-k query term selection, boolean query with term threshold to mitigate these costs and improve the models' efficiency without significantly sacrificing retrieval performance. The results show that Expanded-SPLADE strikes the best balance between effectiveness and efficiency, particularly when handling large datasets. Our findings offer valuable insights for deploying sparse retrieval models in large-scale search engines.