Skip to content

Latest commit

 

History

History
109 lines (95 loc) · 5.23 KB

README.md

File metadata and controls

109 lines (95 loc) · 5.23 KB
Raw

Incorporating Relevance Feedback for Information-Seeking Retrieval using Few-Shot Document Re-Ranking

Paper Badge Slides Badge Slides Badge

This repository contains the code for our EMNLP 2022 paper Incorporating Relevance Feedback for Information-Seeking Retrieval using Few-Shot Document Re-Ranking.

Overview

Pairing a lexical retriever with a neural re-ranking model has set state-of-the-art performance on large-scale information retrieval datasets. This pipeline covers scenarios like question answering or navigational queries, however, for information-seeking scenarios, users often provide information on whether a document is relevant to their query in form of clicks or explicit feedback. Therefore, in this work, we explore how relevance feedback can be directly integrated into neural re-ranking models by adopting few-shot and parameter-efficient learning techniques. Specifically, we introduce a kNN approach that re-ranks documents based on their similarity with the query and the documents the user considers relevant. Further, we explore Cross-Encoder models that we pre-train using meta-learning and subsequently fine-tune for each query, training only on the feedback documents. To evaluate our different integration strategies, we transform four existing information retrieval datasets into the relevance feedback scenario. Extensive experiments demonstrate that integrating relevance feedback directly in neural re-ranking models improves their performance, and fusing lexical ranking with our best performing neural re-ranker outperforms all other methods by 5.2 nDCG@20.

Requirements

  • python 3.10+
  • if docker is availbale, elasticsearch will be run in a container, else the bare metal version will be started.

Setup

Virtual Environment and Dependencies

  1. Setup your python virtual environment, for example: 
    python -m venv .venv
  2. Install dependencies 
    make install
    If you want to contribute to the repository, please also install the dev dependencies: 
    make install-dev

Datasets and Preprocessing

The paths where to store raw, preprocessed and experiments data can be modified in the .env file. Once these are set, run the following commands:

  1. Download Datasets 
    make download
  2. Index corpora in elasticsearch. 
    make index

Experiments

The following commands run the experiments using the default parameters. For the full set of available arguments, see inc_rel/args.py.

2nd Stage Retrieval and Query Expansion

Creating the index the last step in the Setup also runs the experiments over the second stage retrieval. By default, the query is expanded with [4, 8, 16, 32, 64] terms using ElasticSearchs MoreLikeThis query.

kNN

make knn args="--dataset <dataset>"

Zero-Shot

make knn args="--dataset <dataset> --num_samples <num_samples>"

Query Fine-Tuning

make args="--dataset <dataset> --num_samples <num_samples>"

Meta-Learning + Query Fine-Tuning

make args="--dataset <dataset> --num_samples <num_samples>"

Rank-Fusion

make args="--dataset <dataset> --num_samples <num_samples> --result_files <path to first result file> <path to second result file>

Contact

This project is maintained by Tim Baumgärtner.

Citation

If you find this work useful, please considering citing the following paper:

@inproceedings{baumgartner-etal-2022-incorporating,
    title = "Incorporating Relevance Feedback for Information-Seeking Retrieval using Few-Shot Document Re-Ranking",
    author = {Baumg{\"a}rtner, Tim  and
      Ribeiro, Leonardo F. R.  and
      Reimers, Nils  and
      Gurevych, Iryna},
    booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing",
    month = dec,
    year = "2022",
    address = "Abu Dhabi, United Arab Emirates",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.emnlp-main.614",
    pages = "8988--9005",
}

Disclaimer

This repository contains experimental software and is published for the sole purpose of giving additional background details on the respective publication.