Exploring Data-Centric Strategies for French Patent Classification: A Baseline and Comparisons

1712

Authors

nan

Abstract

AbstractThis paper proposes a novel approach to French patent classification leveraging data-centric strategies. We compare different approaches for the two deepest levels of the IPC hierarchy: the IPC group and subgroups. Our experiments show that while simple ensemble strategies work for shallower levels, deeper levels require more sophisticated techniques such as data augmentation, clustering, and negative sampling. Our research highlights the importance of language-specific features and data-centric strategies for accurate and reliable French patent classification. It provides valuable insights and solutions for researchers and practitioners in the field of patent classification, advancing research in French patent classification.

Keywords

nan

Full Content

Exploring Data-Centric Strategies for French Patent Classification: A Baseline and Comparisons You Zuo1,2 Houda Mouzoun3 Samir Ghamri Doudane3 Kim Gerdes1,4 Benoît Sagot2 (1) Qatent, Paris, France (2) Inria, Paris, France (3) Institut national de la propriété industrielle, Courbevoie, France (4) Laboratoire Interdisciplinaire des Sciences du Numérique, Université Paris-Saclay, Orsay, France firstname.lastname@inria.fr, hmouzoun@inpi.fr, sghamridoudane@inpi.fr, gerdes@lisn.fr ABSTRACT This paper proposes a novel approach to French patent classification leveraging data-centric strategies. We compare different approaches for the two deepest levels of the IPC hierarchy : the IPC group and subgroups. Our experiments show that while simple ensemble strategies work for shallower levels, deeper levels require more sophisticated techniques such as data augmentation, clustering, and negative sampling. Our research highlights the importance of language-specific features and data-centric strategies for accurate and reliable French patent classification. It provides valuable insights and solutions for researchers and practitioners in the field of patent classification, advancing research in French patent classification. RÉSUMÉ Exploration des stratégies centrées sur les données pour la classification des brevets français : Une base de référence et des comparaisons Cet article propose une nouvelle approche de classification des brevets français qui s’appuie sur des stratégies centrées sur les données. Nous comparons différentes approches pour les deux niveaux les plus profonds de la hiérarchie IPC : le groupe IPC et les sous-groupes. Nos expériences montrent que les stratégies d’ensemble simples fonctionnent pour les niveaux peu profonds, mais que les niveaux profonds nécessitent des techniques plus sophistiquées telles que l’augmentation de données, le regroupement et l’échantillonnage négatif. Notre recherche met en évidence l’importance des caractéristiques spécifiques à la langue et des stratégies centrées sur les données pour une classification précise et fiable des brevets français. Elle fournit des informations et des solutions précieuses pour les chercheurs et les praticiens dans le domaine de la classification des brevets, en faisant progresser la recherche en classification des brevets en français. KEYWORDS : Patent Classification, Extreme Multi-label Text Classification, Deep Learning. MOTS-CLÉS: Classification de Brevets d’Invention, Classification de Textes Multi-labels Extrême, Apprentissage Profond. Actes de CORIA-TALN 2023 Paris, France, 5 au 9 juin 2023 Actes de la 30e Conférence sur le Traitement Automatique des Langues Naturelles (TALN), volume 1 : travaux de recherche originaux – articles longs, pages 349–365. Cette œuvre est mise à disposition sous licence Attribution 4.0 International. 1 Introduction A patent is a legal document that grants its holder the exclusive right to prevent others from making, using, selling, offering for sale, or importing the patented invention within a certain jurisdiction for a specific period of time. Patent databases are valuable sources of information that reflect global innova- tions and technological developments. The number of patent applications has increased significantly over the past two decades, which can be attributed to various factors such as the increasing importance of technology in society, the role of patents in business valuation, and the globalization of commerce. Due to the large volume of patents and patent documents, patent analysis and management have become complex and time-consuming. Additionally, patents are granted …

Content truncated to 500 words

Labeled Text

AbstractThis paper proposes a novel approach to French patent classification leveraging data-centric strategies. We compare different approaches for the two deepest levels of the IPC hierarchy: the IPC group and subgroups. Our experiments show that while simple ensemble strategies work for shallower levels, deeper levels require more sophisticated techniques such as data augmentation, clustering, and negative sampling. Our research highlights the importance of language-specific features and data-centric strategies for accurate and reliable French patent classification. It provides valuable insights and solutions for researchers and practitioners in the field of patent classification, advancing research in French patent classification.

Publication Details

Paper ID:
1712
Year:
2023
Publication:
TIPSTER
Publication Type:
conf
DOI:
nan