Our core analytics engine involves a Vector Space Model, which compares every word in your focal document to every word in every patent and application — an analysis of over 700,000 dimensions for each document comparison. Applying the most advanced data science algorithms, the PRF engine weights each dimension for relevance, identifying associations that are hard to find and creating a “proximity” measurement to every patent or application.
The scale of our methods is immense – we analyze over 450 terabytes of data in seconds using a distributed, parallelized service built by PRF. Although using the “no shortcut” Vector Space Model approach was economically infeasible just three years ago, PRF makes it affordable today to every company and law firm who need reliable patent analytics.
Based on machine learning and artificial intelligence, we have transcribed over ten million examples of office actions and section 102 rejections using advanced neural networks analysis, covering millions of parameters. In addition to our unique Vector Space model, we have transcribed over ten million examples of office actions and section 102 rejections using an extremely advanced neural network analysis, covering millions of parameters, to provide the most objective predicitive analysis available, based on the most advanced machine learning and artificial intelligence techniques. Our analytics then provide the most accurate relationships between patents available.
Even the most advanced technology, when it is applied in a generic way, will produce poor or misleading results. Most patent analysis companies blindly rely on data from second-hand providers, further introducing error into their analysis. Our data goes through numerous corrections, based on steps validated by domain experts, to get the analysis right. Reliable analytics requires deep expertise about the data being analyzed – and the institutional processes that generate that data. Patent attorneys and scholars at PRF have accumulated years of experience working with the USPTO.
Many IP firms use an ad hoc approach in managing their IP and patent assets – from off-shore search firms, online searching, outside counsel, or services that employ off-the-shelf topic modeling tools, such as Keyword, Semantic Gist Analysis, Semantic Search, Latent Derelict Analysis, Latent Semantic Analysis, Semantic Boolean Analysis, and others. There is little or no transparency – you get buzzwords and “black box” descriptions that have not been validated and are difficult to trust. This mixed approach to gathering your patent data can lead to unpredictable and unreliable results.
Because PRF doesn’t track your activity while processing your analytics, and we don’t retain any data after running a report, your request is private and your data secure. For more information, request a copy of our Security Whitepaper.