Pseudo-relevance feedback has traditionally been implemented as an expensive re-retrieval of documents from the target corpus. In this work, we demonstrate that, for high precision metrics, re-ranking the original feedback set provides nearly identical performance to re-retrieval with significantly lower latency.
bibtex
Copied!
@inproceedings{diaz:crm,
year = {2015},
title = {Condensed List Relevance Models},
series = {ICTIR '15},
publisher = {ACM},
pages = {313--316},
month = {May},
booktitle = {Proceedings of the 2015 International Conference on The Theory of Information Retrieval},
author = {Fernando Diaz},
address = {New York, NY, USA}
}