Dataset Card for "BidCorpus"

How to load the datasets

To load one of the datasets, simply provide the tcepi/bidCorpus argument as the first parameter, followed by the name of the desired dataset, such as bid_corpus_raw.

from datasets import load_dataset
dataset = load_dataset("tcepi/bidCorpus", "bidCorpus_raw")

The csv format version of the datasets is available in the \bidCorpus_csvs folder.

Dataset Summary

The BidCorpus dataset consists of various configurations related to bidding documents. It includes datasets for Named Entity Recognition, Multi-label Classification, Sentence Similarity, and more. Each configuration focuses on different aspects of bidding documents and is designed for specific tasks.

Supported Tasks and Leaderboards

The supported tasks are the following:

Dataset	Source	Sub-domain	Task Type	Classes
bidCorpus_NER_keyphrase	-	Seção Objeto de Editais de Licitação	Named Entity Recognition	4
bidCorpus_gold	-	Seção de Habilitação de Editais de Licitação	Multi-label Classification	7
bidCorpus_object_similarity	-	Seção Objeto de Editais de Licitação	Sentence Similarity	2
bidCorpus_objects_correct_allowed	-	Seção Objeto de Editais de Licitação	Multi-class Classification	4
bidCorpus_objects_type	-	Seção Objeto de Editais de Licitação	Multi-class Classification	4
bidCorpus_qual_model	-	Seção de Habilitação de Editais de Licitação	Multi-label Classification	7
bidCorpus_qual_weak_sup	-	Seção de Habilitação de Editais de Licitação	Multi-label Classification	7
bidCorpus_synthetic	-	Seção de Habilitação de Editais de Licitação	Multi-label Classification	7
bidCorpus_sections_type	-	Seções de Editais de Licitação	Multi-label Classification	5
bidCorpus_raw	-	Seções de Editais de Licitação	n/a	n/a

bidCorpus_NER_keyphrase

This dataset is composed of texts from the "object" section of bidding notices. The dataset is labeled with two types of named entities, following the IOB (Inside-Outside-Beginning) format.

1. Object of the bid: Refers to the item to be acquired or the service to be contracted. The tags can be "B-OBJECT" (beginning of the entity) and "I-OBJECT" (continuation of the entity).
2. Municipality of the managing unit: Indicates the location of the entity responsible for the bid. The tags can be "B-MUNICIPALITY" (beginning of the entity) and "I-MUNICIPALITY" (continuation of the entity).

This dataset is intended for training named entity recognition (NER) models, which are used to automatically identify and classify these entities within the texts. The labeled structure of the dataset facilitates the task of teaching models to distinguish between different types of relevant information in the bidding notices. The dataset follows the IOB format for named entity recognition, with entities labeled as either part of the object of the bid or the municipality of the managing unit.

bidCorpus_gold

This dataset consists of texts from the qualification section of bidding notices. Annotated by experts in public procurement, the dataset is multilabel and contains seven labels that indicate possible signs of fraud in public contracts.

1. Certidão de Protesto: Verification of any protests in the company's name.
2. Certificado de Boas Práticas: Assessment of adherence to recommended practices in the sector.
3. Comprovante de Localização: Confirmation of the company's physical address.
4. Idoneidade Financeira: Analysis of the company's financial health.
5. Integralização de Capital: Verification of the company's capital stock integration.
6. Licença Ambiental: Evaluation of compliance with environmental regulations.
7. Limitação de Atestados: Verification of the minimum and maximum number of certificates required.

This dataset is used for training machine learning models to detect signs of fraud in public procurement processes. The multilabel structure allows the models to learn to identify multiple suspicious characteristics simultaneously, providing a valuable tool for the analysis and prevention of fraud in public contracts.

bidCorpus_object_similarity

This dataset is designed to assess text similarity in the "object" section of bidding notices by comparing pairs of distinct notices. Annotated by experts in public procurement, each entry consists of a pair of "object" sections labeled with:

• 1: The sections are similar.
• 0: The sections are not similar.

The dataset supports tasks such as document comparison, clustering, and retrieval. It provides a valuable resource for training and evaluating models on how effectively they can determine similarities between bidding notices.

The pairs are annotated with expert labels to ensure high-quality data, making this dataset ideal for developing and testing algorithms for text similarity analysis. It helps improve the efficiency and accuracy of managing and analyzing bidding documents.

bidCorpus_objects_correct_allowed

This dataset focuses on two classifications related to the "object" section of bidding notices:

1. Object Classification: Determines whether a section is the "object" section of a bidding notice.
2. Permissivity Classification: Assesses whether the object requires permissivity, meaning whether the contract involves areas such as the purchase of medications, cleaning services, or fuels, which might necessitate a certificate of location and an environmental license from regulatory institutions overseeing these activities.

The dataset provides labels for these classifications to support the analysis of compliance and requirements in bidding documents.

bidCorpus_objects_type

This dataset focuses on classifying the type of procurement found in the "object" section of bidding notices. Specifically, it categorizes the type of product or service being bid on into one of the following categories:

• Consumables: Items that are used up or consumed during use, such as office supplies or food products.
• Permanent Assets: Items with a longer lifespan that are intended for repeated use, such as machinery or equipment.
• Services: Non-tangible activities provided to fulfill a need, such as consulting or maintenance services.
• Engineering Works: Projects related to construction, infrastructure, or other engineering tasks.

The dataset provides labels for these classifications to assist in the analysis and organization of bidding documents, facilitating a better understanding of procurement types and aiding in the efficient management of bidding processes.

bidCorpus_qual_model

This dataset consists of texts from the qualification section of bidding notices and is annotated using a model trained on the original fraud detection dataset. It follows a multilabel format similar to the bidCorpus_gold dataset, with labels indicating possible signs of fraud in public procurement processes.

1. Certidão de Protesto: Verification of any protests in the company's name.
2. Certificado de Boas Práticas: Assessment of adherence to recommended practices in the sector.
3. Comprovante de Localização: Confirmation of the company's physical address.
4. Idoneidade Financeira: Analysis of the company's financial health.
5. Integralização de Capital: Verification of the company's capital stock integration.
6. Licença Ambiental: Evaluation of compliance with environmental regulations.
7. Limitação de Atestados: Verification of the minimum and maximum number of certificates required.

Unlike the expert-annotated previous dataset, this dataset has been annotated by a model trained on that data. This automated process ensures consistency and scalability while utilizing insights from the original expert annotations.

The dataset is intended for training and evaluating machine learning models to detect fraud in public procurement. The automated annotation enhances research and development in fraud detection, aiming to improve the accuracy and efficiency of identifying suspicious activities in bidding notices. Its multilabel structure supports the identification and classification of multiple fraud indicators simultaneously, aiding in the ongoing analysis and prevention of fraudulent practices in public contracts.

bidCorpus_qual_weak_sup

This dataset consists of texts from the qualification section of bidding notices and is annotated using weak supervision techniques, specifically through regular expressions. It follows a multilabel format similar to the bidCorpus_gold dataset, with labels indicating possible signs of fraud in public procurement processes.

1. Certidão de Protesto: Verification of any protests in the company's name.
2. Certificado de Boas Práticas: Assessment of adherence to recommended practices in the sector.
3. Comprovante de Localização: Confirmation of the company's physical address.
4. Idoneidade Financeira: Analysis of the company's financial health.
5. Integralização de Capital: Verification of the company's capital stock integration.
6. Licença Ambiental: Evaluation of compliance with environmental regulations.
7. Limitação de Atestados: Verification of the minimum and maximum number of certificates required.

Unlike the previous expert-annotated dataset, this dataset has been annotated using weak supervision techniques, specifically regular expressions. This approach provides a scalable method for labeling data by applying patterns to identify potential fraud indicators, although it may lack the precision of expert annotations.

The dataset is designed for training and evaluating machine learning models to detect fraud in public procurement. The use of weak supervision through regular expressions facilitates the creation of large annotated datasets, supporting research and development in fraud detection. The multilabel structure allows models to classify multiple fraud indicators simultaneously, improving the efficiency of identifying and preventing fraudulent practices in public contracts.

bidCorpus_synthetic

This dataset consists of texts from the qualification section of bidding notices and is annotated using a model trained on the original fraud detection dataset. It follows a multilabel format similar to the bidCorpus_gold dataset, with labels indicating possible signs of fraud in public procurement processes. This dataset underwent modifications to its keywords by incorporating synonyms to evaluate the model's accuracy in handling words different from those it was previously accustomed to.

1. Certidão de Protesto: Verification of any protests in the company's name.
2. Certificado de Boas Práticas: Assessment of adherence to recommended practices in the sector.
3. Comprovante de Localização: Confirmation of the company's physical address.
4. Idoneidade Financeira: Analysis of the company's financial health.
5. Integralização de Capital: Verification of the company's capital stock integration.
6. Licença Ambiental: Evaluation of compliance with environmental regulations.
7. Limitação de Atestados: Verification of the minimum and maximum number of certificates required.

The dataset is intended for training and evaluating machine learning models to detect fraud in public procurement. Its multilabel structure supports the identification and classification of multiple fraud indicators simultaneously, aiding in the ongoing analysis and prevention of fraudulent practices in public contracts.

bidCorpus_sections_type

This dataset classifies different types of sections in bidding notices. The sections are categorized into the following labels:

• Habilitação: Qualification section, where eligibility criteria and requirements are outlined.
• Julgamento: Evaluation section, detailing the criteria and process for assessing bids.
• Objeto: Object section, specifying the item or service being procured.
• Outros: Other sections that do not fall into the categories above.
• Credenciamento: Accreditation section, where the process for validating and registering vendors is described.

The dataset provides a systematic approach to categorize the various sections found in bidding notices, facilitating better organization and analysis of procurement documents.

bidCorpus_raw

This dataset consists of raw, unlabeled texts from sections of bidding notices. The sections included are:

• Objeto: Describes the item or service being procured.
• Julgamento: Outlines the criteria and process for evaluating bids.
• Credenciamento: Details the procedures for vendor registration and validation.
• Condições de Participação: Specifies the conditions required for participation in the bidding process.
• Habilitação: Provides information on the qualifications and eligibility criteria for bidders.

This dataset offers a collection of unprocessed text from various sections of bidding notices, suitable for tasks such as text analysis, feature extraction, and the development of classification models.

Languages

We considered only datasets in Portuguese.

Dataset Structure

Data Instances

bidCorpus_NER_keyphrase

An example of 'train' looks as follows.

{
  "tokens": ["constitui", "objeto", "do", "presente", "edital", "a", "contratacao", "de", "empresa", "de", "engenharia", "para", "execucao", "da", "obra", "e", "/", "ou", "servico", "de", "elaboracao", "de", "plano", "diretor", "de", "arborizacao", "urbana", "de", "teresina", "-", "pi", ".", "a", "forma", "pela", "qual", "deverao", "ser", "executados", "os", "servicos", "licitados", "e", "as", "diversas", "obrigacoes", "dos", "licitantes", "e", "do", "adjudicatario", "do", "objeto", "desta", "licitacao", "estao", "registradas", "neste", "edital", ",", "no", "termo", "de", "referencia", "e", "minuta", "do", "contrato", "e", "demais", "anexos", "que", ",", "igualmente", ",", "integram", "as", "de", "informacoes", "sobre", "a", "licitacao", "."]
  "ner_tags": [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 0, 1, 2, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0] 
}

bidCorpus_gold

An example of 'train' looks as follows.

{
  "text": ["para se habilitarem ao presente convite, os interessados deverao apresentar os documentos abaixo relacionados, nos termos dos artigos 27 a 31 e 32, paragrafo 1, da lei numero 666/93, atraves de seus representantes, no local, data e horario indicados no preambulo deste edital, em envelope inteiramente fechado, contendo em sua parte externa, alem da razao social e endereco da licitante, os seguintes dizeres: prefeitura municipal de angical ..."]
  "labels": "certidao_protesto": 0, "certificado_boas_praticas": 0, "comprovante_localizacao": 0, "idoneidade_financeira": 0, "integralizado": 0, "licenca_ambiental": 0, "n_min_max_limitacao_atestados": 0
}

bidCorpus_object_similarity

An example of 'train' looks as follows.

{
  "nerObjeto1": ["execucao dos servicos de reforma e ampliacao da escola reunida francisco"],
  "nerObjeto2": ["execucao dos servicos de reforma da escola municipal"],
  "humanScore": 1.0,
  "bertscore_ner": 1
}

bidCorpus_objects_correct_allowed

An example of 'train' looks as follows.

{
  "text": ["A presente licitação tem por objeto, selecionar empresas do ramo pertinente, Fornecimento de Lanches, marmitas para atender necessidade das Secretarias e Programa do Município com entrega parcelada ..."],
  "corretude": 1,
  "permitido": 0
}

bidCorpus_objects_type

An example of 'train' looks as follows.

{
  "text": ["destina - se a presente licitacao a prestacao de servicos de pavimentacao em paralelepipedo, conforme especificacoes e quantidades constantes do anexo <numero> sao ..."],
  "label": 0
}

bidCorpus_qual_model

An example of 'train' looks as follows.

{
  "text": ["regras gerais. 1 os documentos de habilitacao deverao ser enviados concomitantemente com o envio da proposta, conforme item 9 deste edital 2 havendo a necessidade de envio de documentos de habilitacao complementares ..."],
  "certidao_protesto": 0, "certificado_boas_praticas": 0, "comprovante_localizacao": 0, "idoneidade_financeira": 0, "integralizado": 0, "licenca_ambiental": 0, "n_min_max_limitacao_atestados": 0
}

bidCorpus_qual_weak_sup

An example of 'train' looks as follows.

{
  "text": ["os licitantes encaminharao, exclusivamente por meio do sistema, concomitantemente com os documentos de habilitacao. exigidos no edital, proposta com a descricao ..."],
  "certidao_protesto": 0, "certificado_boas_praticas": 0, "comprovante_localizacao": 0, "idoneidade_financeira": 0, "integralizado": 0, "licenca_ambiental": 0, "n_min_max_limitacao_atestados": 0
}

bidCorpus_synthetic

An example of 'train' looks as follows.

{
  "text": ["os licitantes encaminharao, exclusivamente por meio do sistema, concomitantemente com os documentos de habilitacao. exigidos no edital, proposta com a descricao ..."],
  "certidao_protesto": 0, "certificado_boas_praticas": 0, "comprovante_localizacao": 0, "idoneidade_financeira": 0, "integralizado": 0, "licenca_ambiental": 0, "n_min_max_limitacao_atestados": 0
}

bidCorpus_sections_type

An example of 'train' looks as follows.

{
  "text": ["IMPUGNAÇÃO DO ATO CONVOCATÓRIO 5.1 No prazo de até 03 (três) dias úteis, antes da data fixada para abertura da Sessão Pública, qualquer pessoa poderá solicitar esclarecimentos e providências sobre o ato convocatório deste pregão ..."],
  "label": "outros"
}

bidCorpus_raw

An example of 'train' looks as follows.

{
  "ID-LICITACAO": 910809.0,
  "ID-ARQUIVO": 745202022.0,
  "OBJETO": "Artigo 20 Definição do Objeto\n1 – O objeto da licitação deve ser definido pela unidade ...",
  "JULGAMENTO":"Artigo 46 Disposições gerais 1 – As licitações podem adotar os modos de disputa aberto, fechado ou combinado, que deve ...",
  "CONDICAO_PARTICIPACAO": "5.1 - A participação no certame se dará por meio da digitação da senha pessoal e intransferível do representante ...",
  "HABILITACAO": "6.1 - Os proponentes encaminharão, exclusivamente por meio do sistema eletrônico, os documentos de habilitação exigidos no edital, proposta ...",
  "CREDENCIAMENTO": "4.1 - O credenciamento é o nível básico do registro cadastral no SICAF, que permite a participação dos interessados na modalidade licitatória ..."
}

Data Fields

bidCorpus_NER_keyphrase

• tokens: a list of string features (list of tokens in a text).
• ner_tags: a list of classification labels (a list of named entity recognition tags).
  List of NER tags

  `O`, `B-LOCAL`, `I-LOCAL`, `B-OBJETO`, `I-OBJETO`

bidCorpus_gold

• text: a string feature (string of factual paragraphs from the case description).
• certidao_protesto: a 'int64` feature (indicates the presence or absence of a protest certificate).
• certificado_boas_praticas: a 'int64` feature (indicates the presence or absence of a good practices certificate).
• comprovante_localizacao: a 'int64` feature (indicates the presence or absence of a location proof).
• idoneidade_financeira: a 'int64` feature (indicates the presence or absence of financial soundness).
• integralizado: a 'int64` feature (indicates the presence or absence of full completion).
• licenca_ambiental: a 'int64` feature (indicates the presence or absence of an environmental license).
• n_min_max_limitacao_atestados: a 'int64` feature (indicates the presence or absence of limitation of certificates).

bidCorpus_object_similarity

• objeto1: a string feature (first object for comparison).
• nerObjeto1: a string feature (NER tags for the first object).
• objeto2: a string feature (second object for comparison).
• nerObjeto2: a string feature (NER tags for the second object).
• humanScore: a float64 feature (human-provided similarity score).
• nerObjeto1_words: a int64 feature (number of words in the first object with NER tags).
• objeto1_words: a int64 feature (number of words in the first object).
• percentual_words: a float64 feature (percentage of similar words).
• nerObjeto2_words: a 'int64` feature (number of words in the second object with NER tags).
• objeto2_words: a int64 feature (number of words in the second object).
• bertscore_ner: a 'int64` feature (BERT score for NER).
• bertscore_objs: a 'int64` feature (BERT score for objects).

bidCorpus_objects_correct_allowed

• text: a list of string features (list of factual paragraphs from the case description).
• corretude: a list of int64 features (correctness score).
• permitido: a list of int64 features (allowed score).

bidCorpus_objects_type

• text: a list of string features (list of factual paragraphs from the case description).
• label: a list of int64 features (classification labels for object types).

bidCorpus_qual_model

• text: a list of string features (list of factual paragraphs from the case description).
• certidao_protesto: a list of int64 features (presence or absence of protest certificate).
• certificado_boas_praticas: a list of int64 features (presence or absence of good practices certificate).
• comprovante_localizacao: a list of int64 features (presence or absence of location proof).
• idoneidade_financeira: a list of int64 features (presence or absence of financial soundness).
• integralizado: a list of int64 features (presence or absence of full completion).
• licenca_ambiental: a list of int64 features (presence or absence of environmental license).
• n_min_max_limitacao_atestados: a list of int64 features (presence or absence of limitation of certificates).

bidCorpus_qual_weak_sup

• text: a list of string features (list of factual paragraphs from the case description).
• certidao_protesto: a list of int64 features (presence or absence of protest certificate).
• certificado_boas_praticas: a list of int64 features (presence or absence of good practices certificate).
• comprovante_localizacao: a list of int64 features (presence or absence of location proof).
• idoneidade_financeira: a list of int64 features (presence or absence of financial soundness).
• integralizado: a list of int64 features (presence or absence of full completion).
• licenca_ambiental: a list of int64 features (presence or absence of environmental license).
• n_min_max_limitacao_atestados: a list of int64 features (presence or absence of limitation of certificates).

bidCorpus_synthetic

• text: a list of string features (list of factual paragraphs from the case description).
• certidao_protesto: a list of int64 features (presence or absence of protest certificate).
• certificado_boas_praticas: a list of int64 features (presence or absence of good practices certificate).
• comprovante_localizacao: a list of int64 features (presence or absence of location proof).
• idoneidade_financeira: a list of int64 features (presence or absence of financial soundness).
• integralizado: a list of int64 features (presence or absence of full completion).
• licenca_ambiental: a list of int64 features (presence or absence of environmental license).
• n_min_max_limitacao_atestados: a list of int64 features (presence or absence of limitation of certificates).

bidCorpus_sections_type

• text: a list of string features (list of factual paragraphs from the case description).
• label: a list of string features (classification labels for sections types).

bidCorpus_raw

• ID-LICITACAO: a list of float64 features (auction ID).
• ID-ARQUIVO: a list of float64 features (file ID).
• OBJETO: a list of string features (object of the auction).
• JULGAMENTO: a list of string features (judgment details).
• CONDICAO_PARTICIPACAO: a list of string features (participation conditions).
• HABILITACAO: a list of string features (qualification details).
• CREDENCIAMENTO: a list of string features (accreditation details).

Data Splits

Dataset	Training	Development	Test	Total
bidCorpus_NER_keyphrase	1.632	204	204	2.040
bidCorpus_gold	1.454	182	182	1.818
bidCorpus_object_similarity	1.403	175	176	1.754
bidCorpus_objects_correct_allowed	1.089	136	137	1.362
bidCorpus_objects_type	1.709	214	214	2.137
bidCorpus_qual_model	177.133	22.142	22.142	221.417
bidCorpus_qual_weak_sup	177.133	22.142	22.142	221.417
bidCorpus_synthetic	1.454	182	182	1.818
bidCorpus_sections_type	1.224	153	153	1.530

Dataset Creation

Curation Rationale

More Information Needed

Source Data

Initial Data Collection and Normalization

More Information Needed

Who are the source language producers?

More Information Needed

Annotations

Annotation process

More Information Needed

Who are the annotators?

More Information Needed

Personal and Sensitive Information

More Information Needed

Considerations for Using the Data

Social Impact of Dataset

More Information Needed

Discussion of Biases

More Information Needed

Other Known Limitations

More Information Needed

Additional Information

More Information Needed

Dataset Curators

Licensing Information

More Information Needed

Citation Information

Contributions