datalyes/retrieval_IN

In-Domain Patent Retrieval

Dataset Description

This dataset is part of PatenTEB, a comprehensive benchmark for evaluating text embedding models on patent-specific tasks. PatenTEB comprises 15 tasks across retrieval, classification, paraphrase detection, and clustering, with 2.06 million examples designed to reflect real-world patent analysis workflows.

Paper: PatenTEB: A Comprehensive Benchmark and Model Family for Patent Text Embedding

Task Details

• Task Name: retrieval_IN
• Task Type: Retrieval
• Test Samples: 15,809

Symmetric retrieval within the same technological domain (identical IPC3 codes). Tests fine-grained discrimination within a field. Hard negatives are drawn from MIXED category, ensuring the model learns content-based similarity rather than domain matching.

Dataset Structure

This is a retrieval task where models find relevant patents given a query.

Splits:

• test: Query-document pairs for retrieval evaluation

Columns:

• q
• neg
• pos
• pos_link_type
• neg_computed_domain
• objective
• q_text
• pos_text
• neg_text

Data Sample

Below is a 5-row preview of the test set:

q,neg,pos,pos_link_type,neg_computed_domain,objective,q_text,pos_text,neg_text
116-530-185-733-934,174-815-559-259-599,121-188-879-102-184,CITED_BY,FULL_MIX,IN,"method for producing dry fruit chips [SEP] prepare dried fruit chips having excellent palatability, by immersing fruit chips in syrup containing a ...",process for producing sweetened fruit [SEP] a sweetened fruit having a low water activity relative to its moisture content is prepared by bathing a...,ice cream machine having an auxiliary evaporation tank [SEP] an ice cream machine for cooling liquid ice cream into frozen ice cream includes an ev...
026-551-723-824-076,178-157-197-970-685,067-630-631-954-837,CITED_BY,FULL_MIX,IN,acidic drinks containing low-calorie milk [SEP] obtain a low-caloric milk-containing acidic beverage good in flavor and excellent in long-term pres...,vegetable and/or fruit drink composition [SEP] provided is a drink containing vegetable juice and excellent in ease of drinking. a vegetable drink ...,apparatus for use in the formation of a tobacco pouch product. [SEP] an apparatus (110) for use in the formation and filling of a pouched product (...
044-008-763-333-165,114-968-563-053-156,069-688-104-713-833,CITES,PART_MIX,IN,"biological method of analysis for control of quality of nutraceuticals and medications [SEP] medicine, pharmaceutics. invention relates to field of...","method of screening foods for nutraceuticals [SEP] the invention relates to an assay system for screening nutraceuticals, i.e., foods or food subst...","sweetener compositions and foods, beverages, and consumable products made thereof [SEP] provided herein are compositions with enhanced sweetness or..."
181-829-607-211-583,055-380-609-067-006,019-289-029-539-692,CITES,FULL_MIX,IN,"beverages containing noncaloric sweetener and glycerine [SEP] food industry. beverage-type product contains at least one noncaloric sweetener, eryt...","cola beverages [SEP] cola, reduced calorie cola, and diet cola beverage products and methods for making the same are disclosed. the cola beverages ...",preparation of noncarbonated beverage products with improved microbial stability [SEP] the noncarbonated beverage products have a ph of between 5 a...
077-085-434-715-089,162-350-696-599-730,040-768-067-103-194,CITED_BY,PART_MIX,IN,functional antidiabetic drink [SEP] provided is an antidiabetic drink composition containing silkworm extract capable of dropping the blood sugar l...,the composition comprising extract of cinnamomum cassia bark for improving normal flora and enhancing immune response [SEP] the present invention r...,ganglioside relative compound containing s-glycoside [SEP] obtain the title new relative compound effective as a wide range of an anti-influenza vi...

Evaluation Metrics

This task uses NDCG@10 (Normalized Discounted Cumulative Gain at rank 10) as the primary metric. NDCG measures ranking quality by discounting relevance scores by logarithmic position, normalized by the ideal ranking.

Usage

Load Dataset

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("datalyes/{task_name}")

# Access test split
test_data = dataset['test']

Use with Sentence Transformers

from sentence_transformers import SentenceTransformer

# Load a patent-specialized model
model = SentenceTransformer("datalyes/patembed-base")

# Encode patent texts
embeddings = model.encode(test_data['text'])

Integrate with MTEB

This dataset is designed to be integrated with the MTEB (Massive Text Embedding Benchmark) framework. Integration with MTEB is in progress and will be available once the corresponding pull requests are accepted.

Benchmark Context

This dataset is part of a larger benchmark suite:

Benchmark Component	Description
PatenTEB	15 tasks covering retrieval, classification, paraphrase, clustering
Test Data (Released)	319,320 examples across all 15 tasks
Training/Validation Data	1.74 million examples (planned for future release)
Total Dataset Size	2.06 million annotated instances

Note: Currently, only the test split is publicly available. Training and validation data release is planned for a future date.

All 15 Tasks (NEW to MTEB):

• 3 classification tasks: Bloom timing, NLI directionality, IPC3 classification
• 2 clustering tasks: IPC-based, Inventor-based
• 8 retrieval tasks: 3 symmetric (IN/MIXED/OUT domain) + 5 asymmetric (fragment-to-full)
• 2 paraphrase tasks: Problem and solution paraphrase detection

MTEB Integration: Upcoming (PR in progress)

Citation

If you use this dataset, please cite our paper:

@misc{ayaou2025patentebcomprehensivebenchmarkmodel,
      title={PatenTEB: A Comprehensive Benchmark and Model Family for Patent Text Embedding}, 
      author={Iliass Ayaou and Denis Cavallucci},
      year={2025},
      eprint={2510.22264},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2510.22264}
}

License

This dataset is released under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) license.

• You are free to share and adapt the material
• You must give appropriate credit
• You may not use the material for commercial purposes
• If you remix, transform, or build upon the material, you must distribute your contributions under the same license

For full license details, see: https://creativecommons.org/licenses/by-nc-sa/4.0/

Contact

• Authors: Iliass Ayaou, Denis Cavallucci
• Institution: ICUBE Laboratory, INSA Strasbourg
• GitHub: github.com/iliass-y/patenteb
• HuggingFace: huggingface.co/datalyes