README.md

---
configs:
- config_name: ConditionalQA-corpus
  data_files:
  - split: test
    path: ConditionalQA/corpus/*
- config_name: ConditionalQA-corpus_coref
  data_files:
  - split: test
    path: ConditionalQA/corpus_coref/*
- config_name: ConditionalQA-docs
  data_files:
  - split: test
    path: ConditionalQA/docs/*
- config_name: ConditionalQA-keyphrases
  data_files:
  - split: test
    path: ConditionalQA/keyphrases/*
- config_name: ConditionalQA-qrels
  data_files:
  - split: train
    path: ConditionalQA/qrels/train.parquet
  - split: dev
    path: ConditionalQA/qrels/dev.parquet
  - split: test
    path: ConditionalQA/qrels/test.parquet
- config_name: ConditionalQA-queries
  data_files:
  - split: train
    path: ConditionalQA/queries/train.parquet
  - split: dev
    path: ConditionalQA/queries/dev.parquet
  - split: test
    path: ConditionalQA/queries/test.parquet
- config_name: Genomics-corpus
  data_files:
  - split: test
    path: Genomics/corpus/*
- config_name: Genomics-corpus_coref
  data_files:
  - split: test
    path: Genomics/corpus_coref/*
- config_name: Genomics-docs
  data_files:
  - split: test
    path: Genomics/docs/*
- config_name: Genomics-keyphrases
  data_files:
  - split: test
    path: Genomics/keyphrases/*
- config_name: Genomics-qrels
  data_files:
  - split: test
    path: Genomics/qrels/test.parquet
- config_name: Genomics-queries
  data_files:
  - split: test
    path: Genomics/queries/test.parquet
- config_name: MIRACL-corpus
  data_files:
  - split: test
    path: MIRACL/corpus/*
- config_name: MIRACL-corpus_coref
  data_files:
  - split: test
    path: MIRACL/corpus_coref/*
- config_name: MIRACL-docs
  data_files:
  - split: test
    path: MIRACL/docs/*
- config_name: MIRACL-keyphrases
  data_files:
  - split: test
    path: MIRACL/keyphrases/*
- config_name: MIRACL-qrels
  data_files:
  - split: train
    path: MIRACL/qrels/train.parquet
  - split: dev
    path: MIRACL/qrels/dev.parquet
  - split: test
    path: MIRACL/qrels/test.parquet
- config_name: MIRACL-queries
  data_files:
  - split: train
    path: MIRACL/queries/train.parquet
  - split: dev
    path: MIRACL/queries/dev.parquet
  - split: test
    path: MIRACL/queries/test.parquet
- config_name: MSMARCO-corpus
  data_files:
  - split: test
    path: MSMARCO/corpus/*
- config_name: MSMARCO-corpus_coref
  data_files:
  - split: test
    path: MSMARCO/corpus_coref/*
- config_name: MSMARCO-docs
  data_files:
  - split: test
    path: MSMARCO/docs/*
- config_name: MSMARCO-keyphrases
  data_files:
  - split: test
    path: MSMARCO/keyphrases/*
- config_name: MSMARCO-qrels
  data_files:
  - split: train
    path: MSMARCO/qrels/train.parquet
  - split: dev
    path: MSMARCO/qrels/dev.parquet
  - split: test
    path: MSMARCO/qrels/test.parquet
- config_name: MSMARCO-queries
  data_files:
  - split: train
    path: MSMARCO/queries/train.parquet
  - split: dev
    path: MSMARCO/queries/dev.parquet
  - split: test
    path: MSMARCO/queries/test.parquet
- config_name: NaturalQuestions-corpus
  data_files:
  - split: test
    path: NaturalQuestions/corpus/*
- config_name: NaturalQuestions-corpus_coref
  data_files:
  - split: test
    path: NaturalQuestions/corpus_coref/*
- config_name: NaturalQuestions-docs
  data_files:
  - split: test
    path: NaturalQuestions/docs/*
- config_name: NaturalQuestions-keyphrases
  data_files:
  - split: test
    path: NaturalQuestions/keyphrases/*
- config_name: NaturalQuestions-qrels
  data_files:
  - split: dev
    path: NaturalQuestions/qrels/dev.parquet
  - split: test
    path: NaturalQuestions/qrels/test.parquet
- config_name: NaturalQuestions-queries
  data_files:
  - split: dev
    path: NaturalQuestions/queries/dev.parquet
  - split: test
    path: NaturalQuestions/queries/test.parquet
- config_name: nq-hard
  data_files:
  - split: test
    path: NaturalQuestions/nq-hard/*
dataset_info:
  features:
  - name: doc_id
    dtype: string
  - name: title
    dtype: string
  - name: passage_ids
    sequence: string
  - name: passages
    sequence: string
  - name: is_candidate
    sequence: bool
  splits:
  - name: test
    num_bytes: 13421074669
    num_examples: 5758285
  download_size: 7956252663
  dataset_size: 13421074669
---

# DAPR: Document-Aware Passage Retrieval

This datasets repo contains the queries, passages/documents and judgements for the data used in the [DAPR](https://arxiv.org/abs/2305.13915) paper.

DAPR is a benchmark for document-aware passage retrieval: given a (large) collection of documents, relevant passages within these documents for a given query are required to be returned. 

A key focus of DAPR is forcing/encouraging retrieval systems to utilize the document-level context which surrounds the relevant passages. An example is shown below:

<img src='https://raw.githubusercontent.com/UKPLab/acl2024-dapr/main/imgs/motivative-example.png' width='300'>

> In this example, the query asks for a musician or a group who has ever played at a certain venue. However, the gold relevant passage mentions only the reference noun, "the venue" but its actual name, "the Half Moon, Putney". The model thus needs to explore the context from the belonging document of the passage, which in this case means coreference resolution.


## Overview

For the DAPR benchmark, it contains 5 datasets:
| Dataset | #Queries (test) | #Documents | #Passages
| --- |  --- | --- | --- |
| [MS MARCO](https://microsoft.github.io/msmarco/) | 2,722 | 1,359,163 | 2,383,023* |
| [Natural Questions](https://ai.google.com/research/NaturalQuestions) | 3,610 | 108,626 | 2,682,017|
| [MIRACL](https://project-miracl.github.io/) | 799 | 5,758,285 |32,893,221|
| [Genomics](https://dmice.ohsu.edu/trec-gen/) | 62 | 162,259 |12,641,127|
| [ConditionalQA](https://haitian-sun.github.io/conditionalqa/) | 271 | 652 |69,199|

And additionally, NQ-hard, the hard subset of queries from Natural Questions is also included (516 in total). These queries are hard because understanding the document context (e.g. coreference, main topic, multi-hop reasoning, and acronym) is necessary for retrieving the relevant passages.

> Notes: for MS MARCO, its documents do not provide the gold paragraph segmentation and we only segment the document by keeping the judged passages (from the MS MARCO Passage Ranking task) standing out while leaving the rest parts surrounding these passages. These passages are marked by `is_candidate==true`.

> For Natural Questions, the training split is not provided because the duplidate timestamps cannot be compatible with the queries/qrels/corpus format. Please refer to https://public.ukp.informatik.tu-darmstadt.de/kwang/dapr/data/NaturalQuestions/ for the training split.

## Load the dataset
### Loading the passages
One can load the passages like this:
```python
from datasets import load_dataset

dataset_name = "ConditionalQA"
passages = load_dataset("UKPLab/dapr", f"{dataset_name}-corpus", split="test")
for passage in passages:
    passage["_id"]  # passage id
    passage["text"]  # passage text
    passage["title"]  # doc title
    passage["doc_id"]
    passage["paragraph_no"]  # the paragraph number within the document
    passage["total_paragraphs"]  # how many paragraphs/passages in total in the document
    passage["is_candidate"]  # is this passage a candidate for retrieval
```

Or strem the dataset without downloading it beforehand:
```python
from datasets import load_dataset

dataset_name = "ConditionalQA"
passages = load_dataset(
    "UKPLab/dapr", f"{dataset_name}-corpus", split="test", streaming=True
)
for passage in passages:
    passage["_id"]  # passage id
    passage["text"]  # passage text
    passage["title"]  # doc title
    passage["doc_id"]
    passage["paragraph_no"]  # the paragraph number within the document
    passage["total_paragraphs"]  # how many paragraphs/passages in total in the document
    passage["is_candidate"]  # is this passage a candidate for retrieval
```

### Loading the qrels
The qrels split contains the query relevance annotation, i.e., it contains the relevance score for (query, passage) pairs.
```python
from datasets import load_dataset

dataset_name = "ConditionalQA"
qrels = load_dataset("UKPLab/dapr", f"{dataset_name}-qrels", split="test")
for qrel in qrels:
    qrel["query_id"]  # query id (the text is available in ConditionalQA-queries)
    qrel["corpus_id"]  # passage id
    qrel["score"]  # gold judgement

```
We present the NQ-hard dataset in an extended format of the normal qrels with additional columns:
```python
from datasets import load_dataset

qrels = load_dataset("UKPLab/dapr", "nq-hard", split="test")
for qrel in qrels:
    qrel["query_id"]  # query id (the text is available in ConditionalQA-queries)
    qrel["corpus_id"]  # passage id
    qrel["score"]  # gold judgement

    # Additional columns:
    qrel["query"]  # query text
    qrel["text"]  # passage text
    qrel["title"]  # doc title
    qrel["doc_id"]
    qrel["categories"]  # list of categories about this query-passage pair
    qrel["url"]  # url to the document in Wikipedia
```

## Retrieval and Evaluation
The following shows an example, how the dataset can be used to build a semantic search application.
> This example is based on [clddp](https://github.com/kwang2049/clddp/tree/main) (`pip install -U cldpp`). One can further explore this [example](https://github.com/kwang2049/clddp/blob/main/examples/search_fiqa.sh) for convenient multi-GPU exact search.

```python
# Please install cldpp with `pip install -U cldpp`
from clddp.retriever import Retriever, RetrieverConfig, Pooling, SimilarityFunction
from clddp.dm import Separator
from typing import Dict
from clddp.dm import Query, Passage
import torch
import pytrec_eval
import numpy as np
from datasets import load_dataset


# Define the retriever (DRAGON+ from https://arxiv.org/abs/2302.07452)
class DRAGONPlus(Retriever):
    def __init__(self) -> None:
        config = RetrieverConfig(
            query_model_name_or_path="facebook/dragon-plus-query-encoder",
            passage_model_name_or_path="facebook/dragon-plus-context-encoder",
            shared_encoder=False,
            sep=Separator.blank,
            pooling=Pooling.cls,
            similarity_function=SimilarityFunction.dot_product,
            query_max_length=512,
            passage_max_length=512,
        )
        super().__init__(config)


# Load data:
passages = load_dataset("UKPLab/dapr", "ConditionalQA-corpus", split="test")
queries = load_dataset("UKPLab/dapr", "ConditionalQA-queries", split="test")
qrels_rows = load_dataset("UKPLab/dapr", "ConditionalQA-qrels", split="test")
qrels: Dict[str, Dict[str, float]] = {}
for qrel_row in qrels_rows:
    qid = qrel_row["query_id"]
    pid = qrel_row["corpus_id"]
    rel = qrel_row["score"]
    qrels.setdefault(qid, {})
    qrels[qid][pid] = rel

# Encode queries and passages: (refer to https://github.com/kwang2049/clddp/blob/main/examples/search_fiqa.sh for multi-GPU exact search)
retriever = DRAGONPlus()
retriever.eval()
queries = [Query(query_id=query["_id"], text=query["text"]) for query in queries]
passages = [
    Passage(passage_id=passage["_id"], text=passage["text"]) for passage in passages
]
query_embeddings = retriever.encode_queries(queries)
with torch.no_grad():  # Takes around a minute on a V100 GPU
    passage_embeddings, passage_mask = retriever.encode_passages(passages)

# Calculate the similarities and keep top-K:
similarity_scores = torch.matmul(
    query_embeddings, passage_embeddings.t()
)  # (query_num, passage_num)
topk = torch.topk(similarity_scores, k=10)
topk_values: torch.Tensor = topk[0]
topk_indices: torch.LongTensor = topk[1]
topk_value_lists = topk_values.tolist()
topk_index_lists = topk_indices.tolist()

# Run evaluation with pytrec_eval:
retrieval_scores: Dict[str, Dict[str, float]] = {}
for query_i, (values, indices) in enumerate(zip(topk_value_lists, topk_index_lists)):
    query_id = queries[query_i].query_id
    retrieval_scores.setdefault(query_id, {})
    for value, passage_i in zip(values, indices):
        passage_id = passages[passage_i].passage_id
        retrieval_scores[query_id][passage_id] = value
evaluator = pytrec_eval.RelevanceEvaluator(
    query_relevance=qrels, measures=["ndcg_cut_10"]
)
query_performances: Dict[str, Dict[str, float]] = evaluator.evaluate(retrieval_scores)
ndcg = np.mean([score["ndcg_cut_10"] for score in query_performances.values()])
print(ndcg)  # 0.21796083196880855
```

## Note
This dataset was created with `datasets==2.15.0`. Make sure to use this or a newer version of the datasets library.

## Citation
If you use the code/data, feel free to cite our publication [DAPR: A Benchmark on Document-Aware Passage Retrieval](https://arxiv.org/abs/2305.13915):
```bibtex 
@article{wang2023dapr,
    title = "DAPR: A Benchmark on Document-Aware Passage Retrieval",
    author = "Kexin Wang and Nils Reimers and Iryna Gurevych", 
    journal= "arXiv preprint arXiv:2305.13915",
    year = "2023",
    url = "https://arxiv.org/abs/2305.13915",
}
```