Add readme
Browse files
README.md
ADDED
|
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
# Scores of generated queries
|
| 3 |
+
|
| 4 |
+
This repo contains the scores files pertaining to [this study](https://github.com/Watheq9/d2qminus-repro). In particular, we scored the expansion queries generated by [T5-based Doc2Query model](https://huggingface.co/castorini/monot5-base-msmarco) for MSMARCO-v1 passage dataset and a subset of BEIR benchemark.
|
| 5 |
+
We used [ELECTRA](https://huggingface.co/crystina-z/monoELECTRA_LCE_nneg31) cross-encoder to get the relevance scores between the document text and its expansion queries. More details in the study repo [here](https://github.com/Watheq9/d2qminus-repro).
|
| 6 |
+
|
| 7 |
+
## Structure
|
| 8 |
+
All files are .jsonl files with the following three columns per line: ["id", "predicted_queries","querygen_score"].
|
| 9 |
+
So, each file contains the document id, the expansion queries and their corresponding ELECTRA relevance scores.
|
| 10 |
+
Here are the matching of each dataset:
|
| 11 |
+
|
| 12 |
+
`msmarco-v1-80-scored-queries.jsonl` is for MSMarco-v1 dataset.
|
| 13 |
+
|
| 14 |
+
`dbpedia-20-scored-queries.jsonl` is for DBPedia dataset.
|
| 15 |
+
|
| 16 |
+
`quora-20-scored-queries.jsonl` is for Quora dataset.
|
| 17 |
+
|
| 18 |
+
`robust04-20-scored-queries.jsonl` is for Robust04 dataset.
|
| 19 |
+
|
| 20 |
+
`trec-covid-20-scored-queries.jsonl` is for TREC-COVID dataset.
|
| 21 |
+
|
| 22 |
+
`webis-touche2020-20-scored-queries.jsonl` is for Touché-2020 dataset.
|
| 23 |
+
|
| 24 |
+
## Credit
|
| 25 |
+
The N=80 expansion queries of MSMARCO-v1 were copied from this [repository](https://github.com/castorini/docTTTTTquery). Please cite their work.
|
| 26 |
+
|
| 27 |
+
The N=20 expansion queries of BEIR benchemark were copied from this [repository](https://huggingface.co/income). Please cite their work.
|
| 28 |
+
|
| 29 |
+
## Citation
|
| 30 |
+
|
| 31 |
+
If you used any piece of this repository, please consider citing our work:
|
| 32 |
+
|
| 33 |
+
```plaintext
|
| 34 |
+
@inproceedings{mansour2024revisit,
|
| 35 |
+
title={Revisiting Document Expansion and Filtering for Effective First-Stage Retrieval},
|
| 36 |
+
author={Mansour, Watheq and Zhuang, Shengyao and Zhuang, Guido and Mackenzie, Joel},
|
| 37 |
+
booktitle = {Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval},
|
| 38 |
+
year={2024},
|
| 39 |
+
publisher = {Association for Computing Machinery},
|
| 40 |
+
series = {SIGIR '24}
|
| 41 |
+
}
|
| 42 |
+
```
|
| 43 |
+
|
| 44 |
+
|
| 45 |
+
---
|
| 46 |
+
license: cc-by-4.0
|
| 47 |
+
---
|