datalyes/retrieval_OUT

Out-of-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_OUT
• Task Type: Retrieval
• Test Samples: 15,462

Symmetric retrieval across completely disjoint technological domains (disjoint IPC3 codes). Evaluates cross-disciplinary knowledge transfer. Hard negatives from OUT-domain category test true cross-domain relevance discrimination. Performance typically degrades 3-6× compared to IN-domain retrieval due to vocabulary mismatch.

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
023-533-211-045-265,133-381-198-808-894,079-864-038-020-512,CITED_BY,OUT,OUT,"edible oil/fat compositions [SEP] provide an edible oil composition, excellent in digestibility without any oiliness by using a specific amount for...",method and apparatus for engaging a beverage extraction device with a container [SEP] devices and methods for clamping a beverage extraction device...,carbonylation process for the production of aromatic acids and derivatives thereof [SEP] the present invention provides a process for the preparati...
098-040-457-461-766,143-528-752-912-549,030-112-424-801-077,CITED_BY,OUT,OUT,"apparatus for treating food products, controlled as a function of set points [SEP] realize high productivity without any danger in the quality of a...",system and method for determining heat transfer in an environment [SEP] systems and methods are disclosed for determining heat transfer in an envir...,side member for a vehicle [SEP] a side member for a vehicle is formed by an extruded profile as well as a closed profile which are both constructed...
197-047-607-271-820,194-978-359-221-90X,025-606-078-703-931,CITED_BY,OUT,OUT,high-density roasted and powdered coffee having fluidity and its production [SEP] obtain a coffee product having a specific median volume particle ...,bi-modal roller grinder [SEP] a grinder for grinding a material. the grinder may include a number of first rollers for grinding the material to a g...,"multi-actuator hydraulic press [SEP] a hydraulic press; having two platens (3,5), a multiplicity of hydraulic cylinders (2a,2b,2c,2d), capable of a..."
023-533-211-045-265,008-353-361-867-131,141-084-654-411-997,CITED_BY,OUT,OUT,"edible oil/fat compositions [SEP] provide an edible oil composition, excellent in digestibility without any oiliness by using a specific amount for...",method and apparatus for beverage extraction needle force indication [SEP] devices and methods for guiding a needle in movement through a bottle cl...,biasing drive torque to a secondary axle in a motor vehicle powertrain [SEP] a method for transmitting power to the wheels of a motor vehicle inclu...
098-040-457-461-766,000-884-944-503-542,091-017-906-488-308,CITED_BY,OUT,OUT,"apparatus for treating food products, controlled as a function of set points [SEP] realize high productivity without any danger in the quality of a...","food information monitoring system [SEP] systems and methods for gathering data, representative of food condition information, from a plurality of ...","dyed yarn [SEP] a dyed yarn, characterized in that it comprises polytrimethylene terephthalate fibers and exhibits an elastic recovery ratio of 60 ..."

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