Datasets:

Languages:
Portuguese
Size Categories:
1M<n<10M
ArXiv:
License:
eduseiti commited on
Commit
e527905
1 Parent(s): ecab1f8

Adding citation info

Browse files
Files changed (1) hide show
  1. README.md +11 -3
README.md CHANGED
@@ -20,14 +20,22 @@ Quati is licensed under [Creative Commons Attribution 4.0 International (CC BY 4
20
 
21
  ## Citation Information
22
 
23
- <Need to include arxiv link>
24
-
 
 
 
 
 
 
 
 
25
 
26
  ## About Quati and how to use
27
 
28
  Quati dataset is currently released in two versions: one with 1 million passages, and a larger one with 10 million passages. So far we have prepared only validation *qrels* for both versions, annotating 50 topics with an average of 97.78 passages per query on the 10M version, and 38.66 passages per query on the 1M version.
29
 
30
- Quati can be used to evaluate any Information Retrieval system target Brazilian Portuguese Language documents. The dataset creation and annotation pipeline can also be used to further expand the passages annotation, or to create other IR datasets targeting specific Languages. Please refer to our publication for further details about the dataset and its creation process.
31
 
32
 
33
  ### Obtaining the 1M dataset version
 
20
 
21
  ## Citation Information
22
 
23
+ ```
24
+ @misc{bueno2024quati,
25
+ title={Quati: A Brazilian Portuguese Information Retrieval Dataset from Native Speakers},
26
+ author={Mirelle Bueno and Eduardo Seiti de Oliveira and Rodrigo Nogueira and Roberto A. Lotufo and Jayr Alencar Pereira},
27
+ year={2024},
28
+ eprint={2404.06976},
29
+ archivePrefix={arXiv},
30
+ primaryClass={cs.IR}
31
+ }
32
+ ```
33
 
34
  ## About Quati and how to use
35
 
36
  Quati dataset is currently released in two versions: one with 1 million passages, and a larger one with 10 million passages. So far we have prepared only validation *qrels* for both versions, annotating 50 topics with an average of 97.78 passages per query on the 10M version, and 38.66 passages per query on the 1M version.
37
 
38
+ Quati can be used to evaluate any Information Retrieval system target Brazilian Portuguese Language documents. The dataset creation and annotation pipeline can also be used to further expand the passages annotation, or to create other IR datasets targeting specific Languages. Please refer to [our publication](https://arxiv.org/abs/2404.06976) for further details about the dataset and its creation process.
39
 
40
 
41
  ### Obtaining the 1M dataset version