Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

LLM-powered Real-time Patent Citation Recommendation for Financial Technologies

Tianang Deng, Yu Deng, Tianchen Gao, Yonghong Hu, Rui Pan

TL;DR

Financial innovation drives rapid patenting in FinTech, creating a need for timely prior-art discovery. The authors propose a real-time patent citation recommendation framework that combines LLM-based semantic embeddings, HNSW-based approximate nearest-neighbor search, and incremental updating to surface top-k citations for new patents. Using a large CNIPA FinTech patent dataset (428,843 patents, 2000–2024), the approach shows that LLM embeddings improve semantic matching over traditional representations and that incremental HNSW updates yield higher recall with modest overhead. The work demonstrates practical feasibility and potential impact for patent offices and firms by enabling up-to-date citation recommendations that reflect the latest innovations.

Abstract

Rapid financial innovation has been accompanied by a sharp increase in patenting activity, making timely and comprehensive prior-art discovery more difficult. This problem is especially evident in financial technologies, where innovations develop quickly, patent collections grow continuously, and citation recommendation systems must be updated as new applications arrive. Existing patent retrieval and citation recommendation methods typically rely on static indexes or periodic retraining, which limits their ability to operate effectively in such dynamic settings. In this study, we propose a real-time patent citation recommendation framework designed for large and fast-changing financial patent corpora. Using a dataset of 428,843 financial patents granted by the China National Intellectual Property Administration (CNIPA) between 2000 and 2024, we build a three-stage recommendation pipeline. The pipeline uses large language model (LLM) embeddings to represent the semantic content of patent abstracts, applies efficient approximate nearest-neighbor search to construct a manageable candidate set, and ranks candidates by semantic similarity to produce top-k citation recommendations. In addition to improving recommendation accuracy, the proposed framework directly addresses the dynamic nature of patent systems. By using an incremental indexing strategy based on hierarchical navigable small-world (HNSW) graphs, newly issued patents can be added without rebuilding the entire index. A rolling day-by-day update experiment shows that incremental updating improves recall while substantially reducing computational cost compared with rebuild-based indexing. The proposed method also consistently outperforms traditional text-based baselines and alternative nearest-neighbor retrieval approaches.

LLM-powered Real-time Patent Citation Recommendation for Financial Technologies

TL;DR

Financial innovation drives rapid patenting in FinTech, creating a need for timely prior-art discovery. The authors propose a real-time patent citation recommendation framework that combines LLM-based semantic embeddings, HNSW-based approximate nearest-neighbor search, and incremental updating to surface top-k citations for new patents. Using a large CNIPA FinTech patent dataset (428,843 patents, 2000–2024), the approach shows that LLM embeddings improve semantic matching over traditional representations and that incremental HNSW updates yield higher recall with modest overhead. The work demonstrates practical feasibility and potential impact for patent offices and firms by enabling up-to-date citation recommendations that reflect the latest innovations.

Abstract

Rapid financial innovation has been accompanied by a sharp increase in patenting activity, making timely and comprehensive prior-art discovery more difficult. This problem is especially evident in financial technologies, where innovations develop quickly, patent collections grow continuously, and citation recommendation systems must be updated as new applications arrive. Existing patent retrieval and citation recommendation methods typically rely on static indexes or periodic retraining, which limits their ability to operate effectively in such dynamic settings. In this study, we propose a real-time patent citation recommendation framework designed for large and fast-changing financial patent corpora. Using a dataset of 428,843 financial patents granted by the China National Intellectual Property Administration (CNIPA) between 2000 and 2024, we build a three-stage recommendation pipeline. The pipeline uses large language model (LLM) embeddings to represent the semantic content of patent abstracts, applies efficient approximate nearest-neighbor search to construct a manageable candidate set, and ranks candidates by semantic similarity to produce top-k citation recommendations. In addition to improving recommendation accuracy, the proposed framework directly addresses the dynamic nature of patent systems. By using an incremental indexing strategy based on hierarchical navigable small-world (HNSW) graphs, newly issued patents can be added without rebuilding the entire index. A rolling day-by-day update experiment shows that incremental updating improves recall while substantially reducing computational cost compared with rebuild-based indexing. The proposed method also consistently outperforms traditional text-based baselines and alternative nearest-neighbor retrieval approaches.
Paper Structure (14 sections, 3 figures, 5 tables)

This paper contains 14 sections, 3 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Patent Citation Recommendation Approaches
  4. Domain Characteristics for Financial Patents
  5. Data and Descriptive Analysis
  6. A Three-Stage Approach to Patent Citation Recommendation
  7. Text Embedding
  8. Candidate Pool Construction
  9. Candidate Ranking and Recommendation
  10. Results
  11. Model Comparison
  12. An illustrative example
  13. Incremental Updates for Real-Time Citation Recommendation
  14. Conclusion

Figures (3)

  • Figure 1: The annual number of published patents in the field of Finance applied to the CNIPA.
  • Figure 2: The word cloud of keywords in financial patent abstracts.
  • Figure 3: Flowchart of the proposed three-stage HNSW-based framework for patent citation recommendation, with the incremental real-time updating mechanism discussed in Section \ref{['incremental']}.

Theorems & Definitions (2)

  • Remark 1
  • Remark 2