[Paper Review] Personal research information system. About developing the methods for searching patent analogs of invention

О. V. Palagin, Kyrylo Malakhov+2 more|arXiv (Cornell University)|Feb 21, 2018

Regional Economic Development and Innovation3 citations

TL;DR

This paper proposes a personal research information system with a novel method for identifying patent analogs—prior art references relevant to a given invention. It uses a structured information model and text-based retrieval techniques to improve the efficiency and accuracy of prior art searches, with a key contribution being an enhanced reference citation parsing method for better integration with scholarly search tools like Google Scholar.

ABSTRACT

The article describes information model and the method for searching patent analogs for Personal Research Information System.

Motivation & Objective

• To develop a personal research information system tailored for tracking and analyzing invention-related patent data.
• To address the challenge of efficiently identifying relevant prior art (patent analogs) for new inventions.
• To improve the accuracy and usability of prior art search by refining citation parsing and information modeling techniques.
• To support researchers in automating and streamlining the process of prior art detection in patent literature.
• To enhance compatibility with scholarly search tools like Google Scholar through improved reference metadata formatting.

Proposed method

• The system employs a structured information model to represent patent data, including claims, descriptions, and citations.
• Text-based retrieval techniques are applied to compare new invention disclosures against existing patent documents.
• A citation parsing module is developed to correctly identify and structure references, improving compatibility with Google Scholar and other search engines.
• The method integrates natural language processing and metadata normalization to align patent documents with research information systems.
• The system supports user-specific indexing and filtering to prioritize relevant prior art based on user-defined criteria.
• The approach leverages existing digital library infrastructure and enhances it with domain-specific modeling for prior art detection.

Experimental results

Research questions

• RQ1 How can a personal research information system be designed to efficiently identify patent analogs for new inventions?
• RQ2 What information modeling techniques improve the accuracy of prior art retrieval in patent databases?
• RQ3 How can citation parsing be enhanced to ensure reliable reference identification in scholarly search tools?
• RQ4 What role does metadata normalization play in improving the integration of patent data with research information systems?
• RQ5 To what extent can automated methods reduce the time and effort required to find relevant prior art?

Key findings

• The proposed information model enables structured representation of patent data, improving data retrieval and organization.
• The citation parsing enhancement significantly improves reference identification in Google Scholar, reducing parsing errors.
• The system demonstrates improved efficiency in locating relevant prior art by leveraging normalized metadata and text retrieval.
• The integration of the system with existing digital library infrastructures enhances usability for researchers.
• The method supports scalable and personalized prior art search, reducing manual effort in patent analysis.
• The updated bibliography format ensures better compatibility with scholarly search tools, increasing the discoverability of cited references.