Todo list and ideas¶

Agenda¶

Although Patent2Net is fully operational, works fine and is enough to begin using Patent Information, a lot can be done to improve analysis:

Correct the issues (continuous process, of course)
Add some more information in the result html page (ModeleContenuIndex.html). Great to add the treating date (thus can be different from gathering) and P2N version
As information analysis do not allways represent the whole Patent Universe (i.e. french abstrat) provide the proportion of P.U. concerned by each analysis
Treat Designated State(s) information for EP and WO patentes to complete the attractivity maps
Improve the Mindmap option to get it more efficient for creativity (Celso is working on)
Build the entire network as a gephi file for download to let new combined network analysis possible
Use the list of standardised applicant names from EPO to normalize nets and tables. See: [CSV datafile] (http://documents.epo.org/projects/babylon/rawdata.nsf/0/71DE2EB24A084A19C1257F3B0032BA98/)

Add some new capabilities to Patent2Net, i.e.:

Within the Patent Universe, build a drawings gallery with hyperlink to the Espacenet patent (Andre is working on)
Within the Familly Patent Universe, provide all the same analysis as with the Patent Universe (Roberto is working on)
Include the treatment of the Cooperative Patent Classification (CPC) with the proportion of P.U. concerned (http://www.cooperativepatentclassification.org/Archive.html)
Build a small database to display results of a specific (Familly) Patent Universe. Database could be [PouchDB] (https://pouchdb.com/) or equivalent

Provide some new ways of gathering and analysis of patent information, i.e.:

Within the Familly Patent Universe, provide a new range of analysis, considering a familly as a unique entity (invention)
Limit the Familly Patent Universe to the only Priority patents, and provide a complete analysis
Using citations of the Familly Patent Universe, provide genealogic analysis, especially descendants to try to detect invention fronts.
Gather research reports when avalaible and provide analysis chains

New contributions and ideas are always welcome.

[x] Introduce and stabilize new data model and ad-hoc mode
[x] Write about “jq”
[x] Remove attr_object_as_dict in favor of attr.as_dict
[x] Resolve doc.designated_states vs. doc.register.designated_states duplication by providing a dotted name resolver for nested objects
[x] Refactor maps.py and tables.py to p2n.formatter namespace
[x] “p2n adhoc dump –format=raw” mode
[x] Display OPS error message when running invalid queries like “p2n adhoc dump –expression=’pa=grohe and py=2015’”

[o] OPS Register: Always sort event-like data in ascending order?

Right now, sort order is mixed as of “history items” vs. “actions” vs. “{publication,application}_reference”.
[o] Write documentation about data model
[o] Complete implementation of Patent2NetBrevet.from_ops_exchange_document re. citations, equivalents and more
[o] Complete implementation of OPSRegisterDocument
[o] Install Webhook on GitHub for automatic documentation building
[o] Upload pre-release versions to PyPI
[o] Currently, python-epo-ops-client requires to be online because it always attempts to authenticate.

Could this be deferred to the actual first remote access to be able to work completely offline with a prewarmed cache?
[o] Caching improvements
- [o] Increase dogpile cache duration to one year
- [o] Provide api/command to clear the cache
- [o] More fine-grained cache ttl control
[o] Use pyjq for providing built-in filtering, with raw or even named filters.

https://pypi.python.org/pypi/pyjq
[o] Should we compute “register.designated_states - register.countries_lapsed” to determine the actual

list of countries the patent is still valid in?