Technology

Our technology ? It’s magic.

Our algorithms and know-how are highly confidential: they result from years of intense efforts of our highly skilled team. Cloem technology combines human drafting and machine drafting. We use proven patent and claim drafting techniques, according to industry best practices, amplified by leading-edge Natural Language Processing technologies (as well by human crowdsourcing and other undisclosed techniques). These combined technologies provide a highly intensive computational drafting system which demultiply human expertise thanks to the power of machines but which always remains under the control and the direction of a patent expert.

Our model in brief:
- human and machine creation of contents
- industry best practices encoded (e.g. Case Law, claim construction, EPO Examination Guidelines and USPTO M.P.E.P., etc)
- more than 100 proprietary algorithms (and combinations thereof): handling of time, of antecedences, of the order of method steps, etc
- multiple specialized parsers for patent claims
- specialized and proprietary dictionaries (e.g. Wordnet, Wikipedia, derived from the analysis of 70 000 000 patent documents, etc)
- use of synonyms, hyponyms, hyperonyms, holonyms, meronyms (and combinations thereof)
- undisclosed techniques 

Human creation

- patent attorney drafting loops
- crowdsourcing techniques (no disclosure of contents because of fragmentation)
- content creation controls (internal)
- undisclosed techniques

Machine creation

- proprietary and exclusive IP algorithms
- natural language processing algorithms
- semantic technologies
- automated reasoning 
- text mining

  • specialized dictionaries

- over 70’000’000 patent documents analysed (EPO, USPTO)
- highly specialized and proprietary dictionaries, with IPC sub-groups granularity 
- claim construction dictionaries 
- knowledge base of patent standards, litigated patents and key patents
- database of millions of technical facts and common general knowledge
- Internet as a source of vocabulary, including real-time web

  • combinatorial use

(addition, deletion, insertion, replacement, permutation, interversion, negation, substitution,…) of:

- synonyms (e.g. "computer" and "computing device")
- hyponyms (e.g. "mouse" for "input device")
- hyperonyms (e.g. "input device" for "mouse")
- meronyms (e.g. "button" for "mouse")
- holonyms (e.g. "mouse" for "button")
- antonyms (e.g. "close" for "open")

  • more than 100 proprietary claim drafting algorithms (and combinations thereof)

- injection of definitions of terms
- permutations and interversion of methods steps 
- management and reparation of antecedences 
- hardcoding of the claims tree, including forgotten combinations
- claims tree solving (linearization)
- method claims into system claims (and vice versa)
- claim category transcoding
- time management in claims
- structural features handling
- creation of intermediate generalizations, all abstraction levels
- proprietary modifiers (e.g. "releasable", "3D", "touch", etc)
- proprietary implementation of TRIZ
- best claim drafting practices
- …

  • use of multiple dictionaries and ontologies (non-exhaustive)

- Wordnet
- Wikipedia
- widely used dictionaries
- claim construction dictionaries
- reference patent drafting books
- scientific books and articles
- reference books, manuals, handbooks and encyclopedias
- EPO and USPTO Examination Guidelines
- Internet pages / trend detection on technology pages sources (e.g. blogs)
- …

  • controlled generation / meaning creation

- User Interface controls
- proprietary controls (internal)
- IPC (CPC) classification management
- …

  • highly advanced search capabilities

- built-in, centralized and integrated search
- in-depth indexing of data
- visualization tools (soon)