What is the advantage of machine learning supported search over classical keyword search?
If a keyword search is comprehensive, it will typically generate a chronological list of several hundred entries that take a long time to analyse. If a keyword search generates few results, important patents will likely be missed.
With our machine learning approach, queries with up to 20,000 results can be ordered according to machine generated scores related to a specific research field.
The commercially most relevant patents appear at the top of the list while the user can go as far down in relevancy to be comfortable of not having missed anything.
Relevancy scores are calculated based on titles, abstracts, IPC classifications as well as applicant information.
Our machine relevancy rankings are superior compared to established search engines because we carefully define each machine learning model based on thousands of individual patents, based on our hands on knowledge of the energy storage sector.
What is the source of patent information provided by b-science.net?
Our database consists of more than 1.9 Mio. patent documents from the European Patent Office (EPO) database, published in 1980 or later, that either contain the words 'battery' or 'batteries' in the title or abstract, or to which CPC or IPC1-8 patent classification codes H01M (batteries & fuel cells) or H01G (capacitors) were assigned.
Machine learning (ML) credits can be used to instantly machine rank search results on our website, or to create an individual triweekly patent report based on a specific search query. For example, a machine-sorted search query with 2,000 results every 3 weeks will consume 34,000 ML credits per year.
Value-added or sales taxes in the user's home country are not included.
Machine learning credits expire after one year.
We reserve the right to change our prices at any time without prior notice.
Free offerings are provided only once to organizations involved in energy storage research.