Simple Applications of BERT for Ad Hoc Document Retrieval. Yang, W., Zhang, H., & Lin, J. 2019. cite arxiv:1903.10972
Simple Applications of BERT for Ad Hoc Document Retrieval [link]Paper  abstract   bibtex   
Following recent successes in applying BERT to question answering, we explore simple applications to ad hoc document retrieval. This required confronting the challenge posed by documents that are typically longer than the length of input BERT was designed to handle. We address this issue by applying inference on sentences individually, and then aggregating sentence scores to produce document scores. Experiments on TREC microblog and newswire test collections show that our approach is simple yet effective, as we report the highest average precision on these datasets by neural approaches that we are aware of.
@misc{yang2019simple,
  abstract = {Following recent successes in applying BERT to question answering, we explore
simple applications to ad hoc document retrieval. This required confronting the
challenge posed by documents that are typically longer than the length of input
BERT was designed to handle. We address this issue by applying inference on
sentences individually, and then aggregating sentence scores to produce
document scores. Experiments on TREC microblog and newswire test collections
show that our approach is simple yet effective, as we report the highest
average precision on these datasets by neural approaches that we are aware of.},
  added-at = {2021-01-07T16:00:41.000+0100},
  author = {Yang, Wei and Zhang, Haotian and Lin, Jimmy},
  biburl = {https://www.bibsonomy.org/bibtex/213212d282eaeeb0cad0839001b6e48be/festplatte},
  description = {[1903.10972] Simple Applications of BERT for Ad Hoc Document Retrieval},
  interhash = {ef91eca34d259366f25d574233369c33},
  intrahash = {13212d282eaeeb0cad0839001b6e48be},
  keywords = {passage-retrieval bert masterthesis},
  note = {cite arxiv:1903.10972},
  timestamp = {2021-01-07T16:00:41.000+0100},
  title = {Simple Applications of BERT for Ad Hoc Document Retrieval},
  url = {http://arxiv.org/abs/1903.10972},
  year = 2019
}

Downloads: 0