---
license: cc-by-4.0
task_categories:
- text-classification
language:
- en
pretty_name: DFKI Product Corpus
size_categories:
- 1K<n<10K
---
# Dataset Card for DFKI Product Corpus

The DFKI Product Corpus is a dataset of 174 English web pages and social media posts annotated for product and company named entities, and the 
relation CompanyProvidesProduct.

## Dataset Details

### Dataset Description

This repository contains only those sentences from the original full-text documents which contain at least one ORGANIZATION and one PRODUCT entity.
For each ORG-PROD combination in a sentence, the dataset contains an instance, labeled with either "CompanyProvidesProduct" or "no_relation". The dataset
is hence more suited for Relation Extraction, similar to the [TACRED dataset](https://huggingface.co/datasets/DFKI-SLT/tacred), and not for Named Entity Recognition.

- **Curated by:** DFKI-SLT
- **Shared by:** DFKI-SLT
- **Language(s) (NLP):** en-us, en-gb
- **License:** CC-BY-4.0

### Dataset Sources

<!-- Provide the basic links for the dataset. -->

- **Repository:** https://github.com/DFKI-NLP/product-corpus
- **Paper:**  https://aclanthology.org/L18-1704.pdf

## Uses

### Direct Use

The dataset can be used for binary relation classification.


## Dataset Structure

<!-- This section provides a description of the dataset fields, and additional information about the dataset structure such as criteria 
used to create the splits, relationships between data points, etc. -->
The original dataset was annotated at the document level. We used a random 80/10/10 split to create train, dev and test sets. The documents in each split
were then converted to sentence-level JSONL files. The test set documents contained a larger number of ORG and PROD annotations than the dev split, which resulted
in a larger test split in the JSONL version. Sentence splitting and tokenization were performed with Spacy 3.2.0, using the `en_core_web_sm' model. The JSON field
structure follows the structure used in the TACRED dataset and includes fields for 'grammar' and 'type', which are somewhat redundant in our case.
of the JSON follows

Each instance of the dataset contains the following fields:
- id: the instance id of this sentence, a string feature
- tokens: the list of tokens of this sentence, a list feature
- label: the relation label, either "CompanyProvidesProduct" or "no_relation"
- entities: the token-based start and end offsets of the head (ORG) and tail (PROD) entities, a list feature
- grammar: a fixed ["SUBJ", "OBJ"] list, since we do not include inverted relations
- type: a fixed ["ORGANIZATION", "PRODUCT"] list, denoting the entity types of the head and tail entities

## Data Splits

| Train | Validation | Test |
|-------|------------|------|
| 2437  |    330     |  510 |


## Dataset Creation

### Curation Rationale

<!-- Motivation for the creation of this dataset. -->
The dataset was created as part of the PLASS project in the Speech & Language Technology Group of DFKI between 2019-2022. The goal was to create training and evaluation data
for company-product relations expressed in web text, as a means of building a knowledge graph.


### Source Data

<!-- This section describes the source data (e.g. news text and headlines, social media posts, translated sentences, ...). -->
Source data is web text, mainly news, but also product pages, special interest web pages, etc.

#### Data Collection and Processing

<!-- This section describes the data collection and processing process such as data selection criteria, filtering and normalization methods, tools and libraries used, etc. -->

[More Information Needed]

#### Who are the source data producers?

<!-- This section describes the people or systems who originally created the data. It should also include self-reported demographic or identity information for the source data creators if this information is available. -->

[More Information Needed]

### Annotations

<!-- If the dataset contains annotations which are not part of the initial data collection, use this section to describe them. -->
The data contains annotations for Named Entities (ORGANIZATION, PRODUCT) and a single relation type (CompanyProvidesProduct), as well as "no_relation". 

#### Annotation process

<!-- This section describes the annotation process such as annotation tools used in the process, the amount of data annotated, annotation guidelines provided to the annotators, interannotator statistics, annotation validation, etc. -->
See the paper https://aclanthology.org/L18-1704.pdf


#### Who are the annotators?

<!-- This section describes the people or systems who created the annotations. -->

Trained student assistants and researchers at DFKI SLT.

#### Personal and Sensitive Information

<!-- State whether the dataset contains data that might be considered personal, sensitive, or private (e.g., data that reveals addresses, uniquely identifiable names or aliases, racial or ethnic origins, sexual orientations, religious beliefs, political opinions, financial or health data, etc.). If efforts were made to anonymize the data, describe the anonymization process. -->

This dataset contains no personal or sensitive information.

## Bias, Risks, and Limitations

<!-- This section is meant to convey both technical and sociotechnical limitations. -->

[More Information Needed]

### Recommendations

<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->

Users should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.

## Citation

<!-- If there is a paper or blog post introducing the dataset, the APA and Bibtex information for that should go in this section. -->

**BibTeX:**
```
@InProceedings{SCHÖN18.88,
  author = {Saskia Schön and Veselina Mironova and Aleksandra Gabryszak and Leonhard Hennig},
  title = "{A Corpus Study and Annotation Schema for Named Entity Recognition and Relation Extraction of Business Products}",
  booktitle = {Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018)},
  year = {2018},
  month = {May 7-12, 2018},
  address = {Miyazaki, Japan},
  editor = {Nicoletta Calzolari (Conference chair) and Khalid Choukri and Christopher Cieri and Thierry Declerck and Sara Goggi and Koiti Hasida and Hitoshi Isahara and Bente Maegaard and Joseph Mariani and Hélène Mazo and Asuncion Moreno and Jan Odijk and Stelios Piperidis and Takenobu Tokunaga},
  publisher = {European Language Resources Association (ELRA)},
  isbn = {979-10-95546-00-9},
  language = {english}
  }
```

**APA:**
```
Schön, S., Mironova, V., Gabryszak, A., & Hennig, L. (2018). A Corpus Study and Annotation Schema for Named Entity Recognition and Relation Extraction of Business Products. ArXiv, abs/2004.03287.
```

### Contributions

Thanks to [@leonhardhennig](https://github.com/leonhardhennig) for adding this dataset.