Datasets:

Modalities:
Text
Formats:
parquet
Libraries:
Datasets
pandas
File size: 2,933 Bytes
2704031
 
2d9ae44
2704031
 
 
3856fb3
2704031
 
 
3856fb3
2704031
3856fb3
 
2d9ae44
 
 
 
 
 
e8e5130
 
2d9ae44
 
e8e5130
2d9ae44
e8e5130
 
2704031
 
 
 
 
2d9ae44
 
 
 
2704031
461968d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
---
dataset_info:
- config_name: documents
  features:
  - name: chunk
    dtype: string
  - name: chunk_id
    dtype: string
  splits:
  - name: test
    num_bytes: 3161302
    num_examples: 3702
  download_size: 1775726
  dataset_size: 3161302
- config_name: queries
  features:
  - name: chunk_ids
    sequence: string
  - name: query
    dtype: string
  - name: answer
    dtype: string
  splits:
  - name: test
    num_bytes: 19746
    num_examples: 36
  download_size: 14925
  dataset_size: 19746
configs:
- config_name: documents
  data_files:
  - split: test
    path: documents/test-*
- config_name: queries
  data_files:
  - split: test
    path: queries/test-*
---

# ConTEB - ESG Reports

This dataset is part of *ConTEB* (Context-aware Text Embedding Benchmark), designed for evaluating contextual embedding model capabilities. It focuses on the theme of **Industrial ESG Reports**, particularly stemming from the fast-food industry.

## Dataset Summary

This dataset was designed to elicit contextual information. It is built upon [a subset of the ViDoRe Benchmark](https://huggingface.co/datasets/vidore/esg_reports_human_labeled_v2). To build the corpus, we start from the pre-existing collection of ESG Reports, extract the text, and chunk them (using [LangChain](https://github.com/langchain-ai/langchain)'s RecursiveCharacterSplitter with a threshold of 1000 characters). We then manually re-annotate the queries to ensure that they are linked to the relevant chunk of the annotated original page. Queries were manually crafted in the original dataset.

This dataset provides a focused benchmark for contextualized embeddings. It includes a curated set of original documents, chunks stemming from them, and queries.

*   **Number of Documents:** 30 
*   **Number of Chunks:** 3702 
*   **Number of Queries:** 36 
*   **Average Number of Tokens per Doc:** 205.5 

## Dataset Structure (Hugging Face Datasets)
The dataset is structured into the following columns:

*   **`documents`**: Contains chunk information:
    *   `"chunk_id"`:  The ID of the chunk, of the form `doc-id_chunk-id`, where `doc-id` is the ID of the original document and `chunk-id` is the position of the chunk within that document. 
    *   `"chunk"`:  The text of the chunk
*   **`queries`**: Contains query information:
    *   `"query"`: The text of the query.
    *   `"answer"`: The answer relevant to the query, from the original dataset.
    *   `"chunk_ids"`: A list of chunk IDs that are relevant to the query. This is used to link the query to the relevant chunks in the `documents` dataset.

## Usage

We will upload a Quickstart evaluation snippet soon.

## Citation

We will add the corresponding citation soon.

## Acknowledgments

This work is partially supported by [ILLUIN Technology](https://www.illuin.tech/), and by a grant from ANRT France.

## Copyright

All rights are reserved to the original authors of the documents.