How to access AI Classifiers in PatentSight

What are AI Classifiers?

Synchronization of AI Classifiers into Patentsight

How can I use AI Classifiers to perform an analysis?

AI Classifier description and last updated date

Filter search and sort by AI Classifier scores in PatentSight+

See AI Classifier scores for each family in PatentSight+

Cipher Classifiers

 

What are AI Classifiers?

The AI Classifier in PatentSight+ transforms how users explore and analyze patent data by introducing classifier-based filtering and insights.

At its core, an AI Classifier is a supervised machine learning algorithm trained to recognize patent families relevant to a specific technology domain. Each patent family receives a classifier score ranging from 0 to 1, indicating the algorithm’s confidence in its relevance — where 0 means no relevance, and 1 signals a strong match.

This scoring system enables precise, focused analysis of patent portfolios, technology landscapes, and innovation trends. By filtering based on classifier scores, users can quickly hone in on the most relevant inventions, streamline competitive intelligence, and make data-driven decisions with greater confidence.

 

Synchronization of AI Classifiers into PatentSight

To build AI Classifiers and calculate scores in PatentSight+, a subscription to LexisNexis Classification is indeed required. This subscription provides access to the necessary tools and resources for creating and utilizing AI Classifiers effectively. If you're interested in learning more about classifier building or need assistance, please reach out to your Customer Success Manager or contact support@lexisnexisIP.com so we provide detailed information and guidance, tailored to your needs.

 

The synchronization of AI Classifiers into PatentSight+ follows specific rules to ensure that relevant patent families are accurately exported and utilized. Here's a breakdown of the synchronization process:

1. Positive Training or Test Set: Patent families that are part of the positive training or test set are always exported to PatentSight+, regardless of their classifier score. This ensures that all patents identified as relevant during the training phase are included in the analysis.
2. Negative Training or Test Set: Patent families that are part of the negative training or test set are excluded from the export, regardless of their score. This prevents patents deemed irrelevant during training from being included in the analysis.
3. Other Patent Families: For patent families not part of the training or test sets, only those with a classifier score of 0.5 or above are exported to PatentSight+. This threshold ensures that only patents with a moderate to high confidence level of relevance are included, optimizing the quality and focus of the analysis.

These rules help maintain the integrity and accuracy of the patent data used in PatentSight+, allowing users to conduct more precise and meaningful analyses.

 

How can I use AI Classifiers to perform an analysis?

To utilize the AI Classifier in your searches, follow these steps:

• 'Add Search Field': In the 'Field Search' section of your New Workbook upon logging into PatentSight, you'll notice a '+ More' button (as highlighted in the image above). Click this button to add a search field to your search.

• Locate Classifiers: In the pop-up, choose 'AI Classifiers' in the left menu.

• Select 'AI Classifier': Look for the tile labelled 'AI Classifier' and select it.

• Add to Favourites: If you frequently use this field, you can add it to your 'Favourites' by clicking the star icon on the tile. This will make it easily accessible for future searches involving a Classifier.
• Open Filter Object: This action will open another filter object where you can select the 'AI Classifier' for your search.

• Select Options: Choose one or multiple options from the list of AI Classifiers. These options function like technology filters, allowing you to refine and enhance your search based on specific classifier criteria.

By following these steps, you can effectively incorporate AI Classifiers into your search strategy, leveraging their capabilities to focus on the most relevant patent data for your needs.

 

 

An 'AI Classifier' can be added to any search, either individually or in multiples, to act as technology filters. These classifiers help in locating patent families that are related to specific technologies by leveraging the AI Classifier's ability to categorize patents based on their relevance to a particular technology area. For instance, if you are interested in finding patent families related to Virtual Reality technology, you can use the 'AI Classifier' to scan all patent families worldwide.

 

By selecting the appropriate classifier for Virtual Reality, the search will filter and identify patents that are most relevant to this technology area, streamlining the process of finding pertinent patent information. This approach allows for a more targeted and efficient search, ensuring that you focus on the most relevant patents for your research or business objectives.

 

 

An 'AI Classifier' can be added to any search, either individually or in multiples, to act as technology filters. These classifiers help in locating patent families that are related to specific technologies by leveraging the AI Classifier's ability to categorize patents based on their relevance to a particular technology area. 

For instance, if you are interested in finding patent families related to Virtual Reality technology, you can use the 'AI Classifier' to scan all patent families worldwide. By selecting the appropriate classifier for Virtual Reality, the search will filter and identify patents that are most relevant to this technology area, streamlining the process of finding pertinent patent information. 

This approach allows for a more targeted and efficient search, ensuring that you focus on the most relevant patents for your research or business objectives.

 

 

Multiple AI Classifiers can indeed be applied to various portfolios to search for patent families across different technologies, such as AI/ML and Virtual Reality (VR). By using these classifiers, you can organize patent families into specific technology areas within your report. 

This organization allows for a comprehensive analysis of the data, facilitating activities such as competitive intelligence, benchmarking, trend scouting, and portfolio optimization. The familiar environment provided by PatentSight+ ensures that you can efficiently navigate and interpret the data, making informed decisions based on the insights gained from the analysis. 

This approach enhances your ability to understand the competitive landscape, identify emerging trends, and optimize your patent portfolio effectively.

 

In the following search described, the Boolean operator 'AND NOT' is used to refine the search results by excluding specific patent families from the Magic Leap portfolio. 

By applying ‘AND NOT Magic Leap’, you effectively filter out Magic Leap’s portfolio, allowing you to:

• Isolate VR-related patents held by other companies
• Identify unique innovations not associated with Magic Leap
• Conduct sharper competitive benchmarking
• Explore white space opportunities in the VR landscape

This approach enables more targeted insights, helping you focus on what’s happening outside a key competitor’s influence.

 

 

In the next search step, a Technology Matrix chart is created using multiple AI Classifiers and the “multiple owners except” filter. This chart visually categorizes patent portfolios by aligning them with specific technology areas—offering a clear view of who is innovating where.

In this setup:

• The Y-axis represents the AI Classifiers, each corresponding to a different technology domain.
• The X-axis lists the selected patent owners, excluding specific entities for focused analysis.

This matrix allows you to:

• Quickly identify which technology areas are strongly represented across different portfolios.
• Detect technology gaps or strengths across competitors or industries.
• Gain strategic insights into where companies are investing their IP efforts.

By combining AI-driven classification with visual mapping, the Technology Matrix provides a powerful overview of the innovation landscape—filterable, comparable, and actionable.

 

 

 

AI Classifier description and last updated date

For each AI Classifier, a description is typically provided as a tooltip. This description offers valuable information about the classifier, including details about its purpose or the technology area it covers. 

Additionally, the tooltip includes the date of the last modification, which helps users understand when the classifier was last updated or adjusted.

 

Filter search and sort by AI Classifier scores in PatentSight+

To organize and prioritize patents using the AI Classifier score, follow these steps:

1. Access the Patent List: Begin by navigating to the list of patents you wish to organize and prioritize.
2. Locate the Sorting Button: Find the sorting button, which is typically located next to the "Competitive Impact" column, as this is often the default sorting setting.
3. Select AI Classifier Option: Click on the sorting button and choose the "AI Classifier" option from the dropdown menu. This will allow you to sort the patent list based on the AI Classifier scores.
4. Arrange by Relevance: Once the AI Classifier option is selected, the patent list will be rearranged according to the relevance and importance of each patent, as determined by their classifier scores. Higher scores indicate greater relevance to the specified technology area.

By following these steps, you can effectively filter and sort patents to assess their relevance and importance, making it easier to focus on the most pertinent patents for your analysis or decision-making process.

 

 

See AI Classifier scores for each family in PatentSight+

The ‘Result List’ provides a detailed view of AI Classifier scores for each patent family. These scores reflect how strongly a patent family aligns with specific technology areas, as determined by the selected AI Classifiers.

To help improve the quality and accuracy of classifier results, users can provide feedback directly from the Result List:

• Use the ‘Relevant’ button to confirm a correct match.
• Use the ‘Irrelevant’ button to flag results that don't fit.

This feedback is sent to the Administrator, supporting ongoing refinement of classifier performance and ensuring the system stays aligned with real-world insights.

 

AI Classifiers in PatentSight are versatile tools that can significantly enhance the depth and accuracy of your patent analyses across various use cases and search scenarios. 

They allow you to filter and organize patent data based on specific technology areas, making it easier to conduct competitive intelligence, trend scouting, and portfolio optimization. 

If you haven't set up AI Classifiers for the technology areas relevant to your work and are interested in learning more, it's recommended to contact your Customer Success Manager for assistance. 

Additionally, you can reach out directly to support at support@lexisnexisip.com for further support and guidance on setting up and utilizing AI Classifiers effectively.

 

 

 

Cipher Classifiers 

(Note: Cipher Classifiers will be replaced with AI Classifiers after a transition phase)

• In the 'Field Search' section of your New Workbook when you first log into PatentSight, you'll find a '+ More' button (outlined below):

• Select this button. This will then open the 'Add Search field' pop up window. To locate the Cipher Classifiers in your account, select 'custom fields', and then select the tile entitled 'Cipher Classifier'. You can also 'favourite' this field by selecting the star in the tile, so that it is available in your 'Favourites' next time you wish to add a Cipher Classifiers to your search.

• This will then open another window where you can select the 'Cipher Classifier' to be included in your search. I am going to select one, but you can select as many as you choose from the list. Think of them like a technology filter that's added to your search:

• Once you've made your selections, click 'close' and you'll see that the Cipher Classifier has been added to your search. You'll also notice the 'add' button next to the Classifier, making it easy to add further Cipher Classifiers to the search if required:

 

There are many other use cases and search scenarios in which you can use AI Classifiers in PatentSight to enhance your analyses.  If you don't already have AI Classifiers built around the technology areas you're interested in and would like to know more, please reach out to your customer success manager, who will be happy to help you, or send us a note at support@lexisnexisip.com.