Patent full text processing data map
Using the Augmented Reality(AR; AR glass) technology field as an example, we provide an example of the output of patent full text processing using PatentPia GoldenCompass. If you click on the chain(link) marks that are attached to each item in the above utilisation map, the example page will open in a new window.
Composition of patent full text
Patent full text data is largely composed of i) patent bibliography, ii) patent disclosure, and iv) patent drawings.
Patent bibliographic data processing
Right holder data processing
The processing of patent right holder (applicant) data includes i) standardised name resolution, ii) organisation attribute mapping, and iii) address and nationality mapping.
PatentPia's classification of institutional attributes includes i) companies, ii) universities, iii) research organisations, iv) NPEs, and v) individuals. Individuals are when the inventor and applicant have the same title of invention, or there is no assignment. For NPEs, see [Link].
The mapping of addresses and nationalities by right holder is maintained by i) ranking frequency by address, and ii) using reliable non-patent sources such as the web.
In the case of large corporate right holders, they often have R&D organisations in different parts of the world, and the addresses listed in patents applied for in the corporate right holder's name are often not their headquarters, but their respective regions (1 company vs. n addresses in patents applied for in its name).
Also, the company headquarters may be in Country A, but the right holder of the patent is often located in a tax haven, etc., and the address information often includes the address of a subsidiary dedicated to managing the patent. In these cases, mapping to the headquarters address, rather than the address of the tax haven or the managing subsidiary, would be consistent with the context/intent of the data consumer.
Citation(reference)-Forward citation-Patent data processing
The key to processing citation (reference) - forward citation patent data is: i) determining patents (=citation - forward citation relationships between patents) within and across nations, and ii) [citation distinction]. Of course, the latter is preceded by the establishment of national reference patent numbers and foreign reference patent numbers contained in a particular patent in a particular nation.
On the other hand, there are cases where user reference patent numbers are included in the background technology, etc. Currently, PatentPia does not extract patent numbers in this area, as most (virtually all) of the US patents, corresponding to the IDS scheme, are included in the reference data.
Patent families patent data processing
The core of patent families data processing is i) Determining which patents can be included in patent families within and across nations (=establishing patent family relationships between patents), and ii) managing family-related patent sets. Each patent family is assigned a family ID.
Patent classification data processing
The core of patent classification data processing is i) [Maintenance and update of patent classification titles] corresponding to patent classification codes, ii) [Mapping of latest patent classification codes by patent], etc.
Inventor data processing
The core of inventor data processing is i) name maintenance (including last name-first name separation) and identification processing, ii) address and nationality mapping, and iii) ethnicity estimation.
Ethnicity is the estimation of ethnicity (not address) from the expression of an inventor's first name; the ethnicity of a particular inventor is not related to the inventor's address. For examples of ethnicity estimation results, see [Link].
IP law firm data processing
The core of IP law firm data processing is i) name maintenance and identification processing, etc.
Patent disclosure data processing
Field-specific keyword extraction
PatentPia provides the highest level of cleaned keyword data. Keywords are extracted from i) title of invention, ii) abstract, and iii) claims of patents. However, keywords related to purpose-solution-problem are extracted from i) technical field, ii) background technology, and iii) summary.
Claims structuring
Among the claims that make up the composition of a patent claim, the inheritance relationship (determining the parent claim) data is maintained. For each dependent claim, determine the parent claim that the dependent claim cites. Utilising the inheritance/citation relationships between claims, a claim family tree can be generated.
Patent disclosure translation
Translate patent disclosures into major languages.
Process patent drawing data
Drawing file conversion
The file sizes of original patent drawing files vary greatly in resolution. Because of this, patent drawings with very large file sizes provide a bad user experience when viewed on mobile or slow network speeds. Given an original patent drawing file, convert it into n file sizes of drawings, including a standard file size drawing file. n also includes a thumbnail.
Generate figure tags
Patented full text data handling process
The expressions with underline are combined with links to example pages, which you can click to view the specific web contents.