How NGK Insulators is Accelerating Innovation with LexisNexis® Classification 

95% Reduction in Workload by Leveraging AI-powered Patent Classification 

How does a century old company accelerate innovation? Founded in 1919, NGK Insulators is an innovator with deep roots in Japan’s energy and manufacturing ecosystem. The company expanded across three mission-critical sectors: environment, digital society, and energy & industry. It then faced a growing need to analyze diverse patent data at scale. 

Masashi Harada, of NGK’s Intellectual Property Strategy Department, identified a major bottleneck. Conventional search queries were pulling in too many irrelevant patents. This “noise” inflated datasets, consuming significant manual effort to review and filter, delaying insights that were vital to R&D and corporate strategy. This article examines how the company utilized AI-powered patent classification and analytics to streamline its workload (reduction of 95%) and provide reliable insights (classification accuracy of 90%) to its decision makers. 

The challenge: Delays and errors, drawbacks of manual patent classification 

The pain points were acute: 

• Search formulas lacked the specificity needed to target only relevant patents. The results often included patents tied to unrelated applications, such as food-grade membranes or gas separation systems. This was not a representative group of patents for the technologies in which NGK Insulators is involved. 

• Manual review of large datasets, sometimes exceeding 10,000 publications, required up to one month of effort, a significant drain on skilled analysts. 

• The broader the initial search (to avoid missing relevant patents), the more noise was introduced, exacerbating the manual workload. 

• Important strategic meetings were slowed by the time it took to deliver clean insights from patent data. 

The cumulative effect was a high cost in time, accuracy, and opportunity. NGK needed a way to transform its IP function into a proactive engine for innovation, not just a reactive support arm. 

The turning point: Adopting LexisNexis Classification 

In June 2024, the IP team at NGK began working with LexisNexis® Classification, an AI-powered patent classification tool. Built to automate and enhance the precision of patent categorization, the tool enabled Harada’s team to transition from labor-intensive review to intelligent machine learning workflows. This delivered faster and more reliable results. 

Key capabilities that enabled accelerating innovation included: 

• Supervised learning via “training datasets”: Harada’s team trained the system using a curated set of relevant and irrelevant patents. This allowed the AI to internalize nuanced distinctions that traditional Boolean logic couldn’t. 

• The “suggestion” function: This feature surfaced borderline cases. It let analysts validate ambiguous patents and strengthen classifier accuracy from 20% to 85%. 

• Scalable automation: Once trained, the classifier could rapidly apply logic across large patent datasets with minimal human input. This saved thousands of manual review hours. 

• High recall and precision scoring: With metrics of 97% recall and 91% precision, the tool consistently delivered cleaner datasets with confidence. 

These combined features helped NGK reduce the time required to classify and clean a 10,000-patent dataset from 20 workdays to just one, freeing the IP team to focus on insight generation and strategic alignment. 

Masashi Harada, IP Strategy Department, NGK Insulators 

“What took 20 days can now be completed in just one.“

Building a smarter dataset: From queries to training sets 

The journey to accelerating innovation at NGK began with the creation of a “training dataset”, a labeled training set containing 100–150 positive and negative patent examples. Using the LexisNexis Classification “Suggestion” function, Harada iteratively refined the training model, raising the classification strength from 20% to 85% in a single day. 

This enabled automated, large-scale filtering of relevant patents, cutting through the noise and freeing up valuable analyst time for higher-order tasks, such as trend forecasting and competitive mapping. 

Accelerating innovation: Speed, precision, and strategic clarity

Harada’s team applied Classification to a dataset of over 4,700 patents in the automotive ceramics domain. The performance metrics spoke volumes: 

• Recall: 97% (few false negatives) 
• Precision: 91% (very low noise)  

The output was benchmarked against AI models used by the Japan Patent Office and found to perform at least as well, and in some cases even better. NGK was able to isolate key players, identify relevant IP faster, and align its internal stakeholders with accurate insights. 

Using traditional methods, reviewing 10,000 patents could take nearly a month. With Classification, that process now takes a day. The team at NGK realized: 

• Time savings of over 90% 
• Lower labor costs in data cleaning and preparation 
• Increased throughput for strategic analyses 

AI-enabled patent analytics fosters a forward-looking IP strategy 

As NGK pursues its long-term vision to lead in carbon neutrality and digital infrastructure, it’s increasingly clear that traditional solutions are not enough. Harada’s success with Classification marked a turning point, but it’s only one part of a broader transformation. 

To further enhance the precision and efficiency of its analyses, NGK plans to continue leveraging both Classification and LexisNexis^® PatentSight+™  as part of its intellectual property strategy. 

Together, they create a powerful feedback loop:

• Classification ensures that only the most relevant patents enter the analysis pipeline by eliminating noise with precision and speed. 
• LexisNexis PatentSight+ then enables NGK to map competitive landscapes, benchmark innovation strength, monitor white space, and assess portfolio quality, all using high-confidence data. 

By feeding Classification-processed datasets into PatentSight+, NGK benefits from:

• Portfolio benchmarking: Measuring internal R&D output against top competitors in key technologies. 
• Technology scouting: Identifying emerging players and acquisition targets before they hit mainstream radar. 
• Strategic alignment: Visualizing how IP investments support broader corporate initiatives, such as electrification, hydrogen fuel, or AI-based systems. 
• Real-time decision support: Using dashboards and Patent Asset Index scores to brief executives before M&A talks or board meetings. 

For a company operating across fast-changing sectors like energy infrastructure, emissions control, and digital ceramics, accelerating its innovation pipeline is a differentiator. It’s what turns a traditional IP team into a strategic enabler of innovation. 

Masashi Harada, IP Strategy Department, NGK Insulators

“With tools like Classification and PatentSight+, we can efficiently remove noise and perform analyses much faster, reducing workload and improving accuracy, while also achieving significant time and cost savings.” 

About NGK Insulators 

With roots in the Morimura Group and a legacy of innovation in high-voltage insulators, NGK has grown into a leading manufacturer with over ¥1 trillion in combined group revenues. Its strategic focus on energy, environment, and digital solutions positions it at the heart of the global transformation toward sustainability and resilience.