Supporting Vendor

PQAI - Patent Quality Through Artificial Intelligence

PQAI - Patent Quality Through Artificial Intelligence
Summary
Provider:

Georgia IP Alliance, GreyB, Perception Partners, FoundersLegal, InspireIP

Description:

Making Prior art search accessible to all

PQAI is a collaborative open source initiative to build a shared AI-based prior-art search tool. At PQAI, we envision a prior art search engine that is:

  1. Accessible by all, no need for extensive training or funds
  2. Powered by AI to provide extremely effective search results
  3. Capable of searching combinational prior art (103), the most frequent reason for rejection at the USPTO
  4. Supports natural language queries
  5. Simple implementation using a distributed architecture eliminating the need to create and maintain a large database.

The vision is big and it can’t be realized without community  support from across the globe. Here is how you can help support the project:

  1. Advisors - Patent professionals who can test the tool, share their feedback and help PQAI improve and promote the adoption of PQAI.
  2. NLP practitioners - The plan is to make PQAI open source just as LINUX, developers can branch out the code, create different versions. For example: Currently PQAI cannot process the data related to chemistry patents, maybe one of the developers can build a version that caters specifically to chemical patents.
  3. Patent Offices - Patent offices can contribute and benefit by providing PQAI access to their examiners and inventors to help grow the user base and If all the patent offices can have an instance of PQAI search engine and they become a part of the distributed network; the amount of data in which the prior art search shall happen will be humongous, almost universal.

If you wish to participate in this initiative, kindly write to sam@projectpq.ai

Sam Zellner, a prolific inventor, retired executive director innovation at AT&T is the product lead for PQAI.

Categories:
Fee:

Free - Not for profit

Website:
https://projectpq.ai
Included Data
Data:

21.5 million documents (patent + research papers)

Type:

Natural language processing based prior art search

Methods Used
Methods:

Deep learning

Supervised machine learning

Representation learning

Named entity recognition

Neural language models

Learning to rank models

Value Added Data
Data:

N/A

Interface
API Available:

Yes, API access is planned. Details will be made available on PQAI website.

API Link:
Analysis Included
Details:

Feature-by-feature mapping of query with the prior-art references

Data Export
Details:

Top-10 search results can be exported in a PDF or Excel format

File Support
File 1:
File 2:
File 3:
File 4:
File 5:
No items found.

Sign Up for the Newsletter

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

©Copyright ML4Patents | Powered By Patinformatics