Neural document ranking approaches, specifically transformer models, have achieved impressive gains in ranking performance. However, query processing using such over-parameterized models is both resource and time intensive. In this paper, we propose the Fast-Forward index – a simple vector forward index that facilitates ranking documents using interpolation of lexical and semantic scores – as a replacement for contextual re-rankers and dense indexes based on nearest neighbor search. Fast-Forward indexes rely on efficient sparse models for retrieval and merely look up pre-computed dense transformer-based vector representations of documents and passages in constant time for fast CPU-based semantic similarity computation during query processing. We propose index pruning and theoretically grounded early stopping techniques to improve the query processing throughput. We conduct extensive large-scale experiments on TREC-DL datasets and show improvements over hybrid indexes in performance and query processing efficiency using only CPUs. Fast-Forward indexes can provide superior ranking performance using interpolation due to the complementary benefits of lexical and semantic similarities.
%0 Conference Paper
%1 10.1145/3485447.3511955
%A Leonhardt, Jurek
%A Rudra, Koustav
%A Khosla, Megha
%A Anand, Abhijit
%A Anand, Avishek
%B Proceedings of the ACM Web Conference 2022
%C New York, NY, USA
%D 2022
%I Association for Computing Machinery
%K efficiency myown neural-ranking
%P 266–276
%R 10.1145/3485447.3511955
%T Efficient Neural Ranking Using Forward Indexes
%U https://doi.org/10.1145/3485447.3511955
%X Neural document ranking approaches, specifically transformer models, have achieved impressive gains in ranking performance. However, query processing using such over-parameterized models is both resource and time intensive. In this paper, we propose the Fast-Forward index – a simple vector forward index that facilitates ranking documents using interpolation of lexical and semantic scores – as a replacement for contextual re-rankers and dense indexes based on nearest neighbor search. Fast-Forward indexes rely on efficient sparse models for retrieval and merely look up pre-computed dense transformer-based vector representations of documents and passages in constant time for fast CPU-based semantic similarity computation during query processing. We propose index pruning and theoretically grounded early stopping techniques to improve the query processing throughput. We conduct extensive large-scale experiments on TREC-DL datasets and show improvements over hybrid indexes in performance and query processing efficiency using only CPUs. Fast-Forward indexes can provide superior ranking performance using interpolation due to the complementary benefits of lexical and semantic similarities.
%@ 9781450390965
@inproceedings{10.1145/3485447.3511955,
abstract = {Neural document ranking approaches, specifically transformer models, have achieved impressive gains in ranking performance. However, query processing using such over-parameterized models is both resource and time intensive. In this paper, we propose the Fast-Forward index – a simple vector forward index that facilitates ranking documents using interpolation of lexical and semantic scores – as a replacement for contextual re-rankers and dense indexes based on nearest neighbor search. Fast-Forward indexes rely on efficient sparse models for retrieval and merely look up pre-computed dense transformer-based vector representations of documents and passages in constant time for fast CPU-based semantic similarity computation during query processing. We propose index pruning and theoretically grounded early stopping techniques to improve the query processing throughput. We conduct extensive large-scale experiments on TREC-DL datasets and show improvements over hybrid indexes in performance and query processing efficiency using only CPUs. Fast-Forward indexes can provide superior ranking performance using interpolation due to the complementary benefits of lexical and semantic similarities.},
added-at = {2022-11-08T08:50:44.000+0100},
address = {New York, NY, USA},
author = {Leonhardt, Jurek and Rudra, Koustav and Khosla, Megha and Anand, Abhijit and Anand, Avishek},
biburl = {https://www.bibsonomy.org/bibtex/2dd66fdd71a1c0beef5a06b311ab31151/leonhardt},
booktitle = {Proceedings of the ACM Web Conference 2022},
doi = {10.1145/3485447.3511955},
interhash = {9a1c4af4b6c3ba3e3b29a9b451a54188},
intrahash = {dd66fdd71a1c0beef5a06b311ab31151},
isbn = {9781450390965},
keywords = {efficiency myown neural-ranking},
location = {Virtual Event, Lyon, France},
numpages = {11},
pages = {266–276},
publisher = {Association for Computing Machinery},
series = {WWW '22},
timestamp = {2022-11-08T08:50:44.000+0100},
title = {Efficient Neural Ranking Using Forward Indexes},
url = {https://doi.org/10.1145/3485447.3511955},
year = 2022
}