Automating Claims Drafting for IP

Automating the generation of patent specifications from claims that are prepared by humans.

Source

Patent drafting quality check system using natural language algorithms to improve the quality of patent applications. The system reviews and evaluates patent drafts for issues like fluency, typos, and deduplication. It uses a language model trained on patents to analyze drafts and generate a report with suggestions for improvement. The system can also prioritize draft acquisition methods based on historical review scores.

Source

A method for generating patent applications using a dynamic user interface and natural language processing. The user interface features fillable objects where users can drop phrases from claims. These phrases are analyzed to determine their type (e.g., adjective or verb phrase) and context. The system converts these phrases into sentences with proper subjects and context. Additionally, the phrases can be matched with related descriptions in drawings or other documents to enhance accuracy. This approach streamlines the process of creating patent applications by converting claim phrases into coherent and contextually accurate content.

Source

Automated method for generating patent application documents using natural language processing (NLP) and a dynamic user interface. The method involves breaking down a patent claim into regions, analyzing each region with NLP, identifying phrases, determining type (descriptive vs functional), and creating complete sentences from the phrases. This allows converting any type of claim into a coherent document using NLP disambiguation of phrases like functional vs descriptive. The system also identifies immaterial tokens and subjects, and can handle leading phrases with context. The user interface allows editing and modifying phrases.

Source

Patent document creation support system that assists users in creating patent applications by providing automated assistance to fill out the application. The system extracts relevant sections from existing patents and allows the user to select and add them to their own application. It also identifies missing sections and suggests appropriate language for inclusion. The system leverages existing patent documents to provide guidance and structure for the application process.

Source

Patent sentence generation system that allows users to easily create high-quality patent applications by providing a guided process for generating patent specification text based on prior art documents. The system extracts documents similar to the invention claims, presents them to the user, allows selection of one, and generates the patent specification text using words from the selected document. This keeps specification quality consistent compared to manual drafting. The system also has a proofreading feature to improve sentence accuracy.

Source

Generating accurate text for patent applications by combining natural language processing (NLP) with structured templating. The system uses natural language understanding (NLU) to parse input text and extract structured components. It then employs natural language generation (NLG) to produce precise output text using a templating language enhanced with functions to manipulate text based on the data model. This method allows the system to generate sections of patent applications by extracting claim structures from existing applications and inserting the parts into a claim model data object, ensuring accurate and contextually relevant content.

Source

Automated document creator system that automatically creates templates for patent applications based on identified prior art references. The system automatically creates a template for a drafter based on a record of invention form filled out by the inventor, practitioner, or a combination of the two and at least one piece of prior art identified in a search report.

Source

Generating patent applications using natural language processing (NLP) and a dynamic user interface that allows users to rearrange and supplement the output from NLP to improve the quality of the generated patent documents. The system uses NLP to analyze the claims and generate a draft patent application. However, the NLP alone may not produce optimal results due to limitations in understanding the claims' formal structure and disambiguating conditions in actions. The user interface allows the user to reorder and add text to the generated application to address these limitations.

Source

A tool to assist in reading patent documents more accurately, quickly, and comfortably by automatically obtaining and displaying relevant guidance content for drawings, claim elements, and other components of the patent. It displays the illustration data like drawings with element numbers alongside the corresponding guidance targets like claim elements or descriptions. The tool uses algorithms to link the illustration data to specific guidance content sections like claim text or descriptions.

Source

Automated system for generating patent application specifications that aims to improve the quality and consistency of patent applications by learning correspondences between patent claims, patent description, and prior art descriptions. The system accepts new claims and generates corresponding sentences for the rest of the specification based on the learned correspondences. It learns the correspondences by analyzing existing patent documents, claims, descriptions, and prior art. This allows the system to generate cohesive and complete patent applications with a consistent level of quality.

Source

A computer-assisted method for improving the quality of patent and utility model applications by providing word suggestions for claims. The method involves storing pairs of original and corrected words from previous claims. When creating new claims, the system outputs suggestions based on the stored pairs associated with the input words. This provides the patent writer with alternative word choices that have been previously approved and used in similar contexts. This aims to improve the clarity and accuracy of the patent claims by suggesting more appropriate and less ambiguous wording choices.

Source

A patent preparation system that captures important information about the relationship among claim limitations to improve patent drafting efficiency. The system allows users to indicate when a dependent claim corresponds to a specific limitation of an independent claim, rather than the entire claim. It stores this relationship data and generates patent application text that reflects the claim structure. This way, claims, drawings, and specifications can be automatically generated and modified with proper claim dependencies to speed up and improve patent preparation.

Source

Computer-based patent preparation that facilitates efficient preparation of patent documents. The preparation includes identifying a set of invention features, generating a first tree structure, identifying a second tree structure, and automatically generating one or more patent application documents based at least in part on the first tree structure and the second tree structure.

Source

Automatically generating patent application specifications using machine learning to improve consistency and quality. The method involves learning correspondences between patent claims, document descriptions, and prior art. New claims are accepted and sentences other than the claim are generated based on the learned correspondences and the new claim. This allows generating consistent specs with a constant quality level compared to manually written ones.

Source

Automatically creating patent applications using AI that follows a document plan inferred from example patents. The system analyzes existing patents to identify common formatting features and alignment between claims and specifications. When a new set of claims is received, it uses natural language generation to create a patent application consistent with the learned document plan and claims structure. This aims to produce applications that closely match the format and organization of typical patents.

Source

System for enhancing computer-generated patent applications by providing non-explicit support for individual claim features to make them more robust and comprehensive. It does this by using machine learning to identify and extract explicit and non-explicit support from the patent specifications for aligning to the claims. It then uses this extracted data to train a model that can generate new non-explicit support statements for previously unseen claim features. The generated support statements can be inserted into draft patent applications to improve their quality and completeness.

Source

Generating a quality patent application that minimizes associated costs, and more affordable than the current state of the art by producing more patent applications in a lesser amount of time, while still upholding the standards of quality and integrity. The generation includes receiving information relevant to drafting a document, receiving guidelines and standards of patentability, analyzing the received information, determining the patentability of the invention, based on the analyzed received information, and generating the document for the invention, based on a learned behavior.

Source

Streamlining the patent application drafting process using artificial intelligence (AI) to analyze user input, prior art, and patentability guidelines to automatically generate patent applications. The AI learns the behavior of patentable inventions by analyzing existing patents and guidelines. It then uses that knowledge to draft patent applications based on user input and prior art.

Source

Computer method for automatically acquiring and correlating reference numerals of element-nouns in patent claims, specifications, and drawings. The method involves searching the specification to find specification-element-noun reference numerals and connecting them to claim-element-nouns. It also maps drawing-nouns to the correlated specification-element-nouns and claim-element-noun reference numerals. This allows a reader to click on any element-noun in any section to display all related element-nouns and their reference numerals from the other sections.

Source