Intellectual Property Landscape Mapping: A Comprehensive Guide

General information interaction method for a technology transfer office that enables efficient retrieval, viewing, evaluation, analysis, and recommendation of patents to support technology transfer. The method involves forming a data processing instruction based on user input, querying the required patent information, processing it to obtain specific data, and then interacting with the user using the processed results. The processing steps include feature extraction, filtering, and analysis techniques like time sequence analysis, semantic clustering, and fuzzy matching. It leverages techniques like patent network modeling, bidirectional value evaluation, and character string filtering to provide enhanced patent analysis capabilities.

Source

Patent evaluation method and system using natural language processing and complex network algorithms to objectively evaluate patent quality by comparing patents with technologies in the global same industry. The method involves collecting patent documents, generating technical points and clusters from the documents using NLP, and comparing patent clusters to assess the depth and breadth of technology, predict patent lifespan, and judge R&D strength. It leverages NLP to aggregate patents into clusters based on their technical attributes and then compares cluster similarity to evaluate patent quality.

Source

Automatic generation of patent citation landscapes for comprehensive analysis of patent family connections and citations to identify problematic patents and analyze portfolio value. The system identifies related patents in a family, and searches and aggregates backward citations for analysis.

Source

An AI system for tracking and identifying innovations in specific categories across multiple data sources using natural language processing. The system receives user queries with keywords related to a category, extracts relevant information from sources using NLP, and identifies innovations based on qualifying terms like "new," "improved," etc. It can search public websites and databases as well as private sources.

Source

Machine learning-based prediction of future innovation at the company level using big data and predictive analysis techniques. The method involves collecting patent, financial, news, and social media data for predetermined companies over a period, feature extraction, and prediction using machine learning models like logistic regression, naive Bayes, neural networks, support vector machines, and deep belief networks. The goal is to predict if a company will innovate in the future based on historical data.

Source

A method and system to predict technology trends using natural language processing and time series analysis on large volumes of scientific papers. The method involves extracting technical features from the papers using NLP, creating a high-dimensional feature space, and then analyzing the relationships between technologies over time using time series techniques to predict their development trajectories. This allows objective analysis and prediction of technology trends without relying on expert opinions.

Source

Automatic classification of patents into industries using a transductive learning method that leverages patent text and classification information to minimize manual labeling. The method involves determining a set of target patents to classify, generating vectors representing the abstract, claims, and description sections of each patent, concatenating them into a hierarchical vector, computing similarities between patent vectors using an adjacency matrix, and iteratively updating the matrix to converge on accurate industry classifications. The method uses patent text structure, IPC codes, and hierarchical vector representation to extract full patent context for classification.

Source

Automatically reviewing and analyzing vendor assurance reports like SOC reports using natural language processing to extract exceptions, generate summaries, identify trends, and present visualized insights. The system trains a natural language engine on report data to detect exceptions, generate summaries, and identify topics. It extracts exceptions from new reports, summarizes them, finds trends across reports, and presents visualized insights.

Source

Electronically mining intellectual property by using text mining techniques to analyze enterprise documents and industry trends to identify correlations that lead to potential intellectual property. The process involves identifying an industry trend, extracting keywords from enterprise documents, determining relevance scores, optimizing weighting schemes, and ranking documents based on intellectual property potential.

Source

Automated method for analyzing documents related to technology fields to predict the success potential of products, services, companies, etc. The analysis is based on computing coefficients from a patent landscape around the specific aspects of interest, and then weighting and combining those coefficients to calculate a probability score. The coefficients are derived from metrics like citation counts, recency, and relevance. This automated scoring system can help investors, managers, and planners assess the breakthrough potential of technology areas without relying solely on human opinions.

Source

Method for identifying potential white space opportunities in patent analysis using a process that incorporates domain expertise and analysis techniques to uncover hidden areas of technological innovation. The method involves constructing taxonomies based on patents related to a specific subject matter, using keywords to analyze documents, and comparing the taxonomies to identify white space areas with low patent density. The taxonomies are created by partitioning documents into categories based on words near keywords, and refining them with domain expert input.

Source