Unified’s HEVC landscape uses a precision-recall machine learning algorithm to the train the semantic indexing model using known Standard Essential and Patent Pool verified patents. The landscape uses input data such as time ranges, IPC/CPC ranges, inventors, applicants, assignees, and SEP declarations to generate a universe of over 2 million patents with more than 500,000 patent families and over 6,000 self-declared standard essential patents.
The model then uses a number of weighted variables to generate an Alpha Value which measures the similarity between the patent claims and the HEVC standard documents for each patent in the defined universe. Alpha Values are also assigned for all self-declared standard essential patents identified in standard setting organization documents and all major patent pools.
Comprehensive HEVC data:
- Utilizes advanced linguistic and machine-learning algorithms (beyond simple keyword searches) to identify and analyze HEVC-related patents and applications.
- Synthesizes data from various sources into a single database, such as patent specification cross-references from patent pools as well as JCT-VC document repo.
- Includes non-public participants and SSO blanket declarants who have not joined any pools.
- Breakdown results by specific technology specifications to discover hidden insights.
- Get full flexibility to drive the analysis based on user needs and leverage precision text mining tools to speed up the process of analysis.
- Search by company or using a list of patents to assess entire portfolios in seconds.
- Provide industry standard analysis to research & development, licensing teams and management.
- Enables true FRAND negotiation by including undeclared & SEP patents.
- A.I. – Semantic based to ensure non-discriminatory results.
- Methodology developed to be used in litigation.
- Employs statistical sampling to ensure reliability to an acceptable degree (supported by HEVC experts).