Patent · US Active

Metadata-based API attribute extraction

US10579372B1 · kind B1 · utility

8Cited by
0References
20Claims
0Family size

Assignee

Inventors

Key dates

Filing dateDec 8, 2018
Grant dateMar 3, 2020
Priority date
Expiry dateDec 8, 2038

Classification

  • Technology area (CPC G)Physics
  • CPC primaryG06V30/416
  • WIPO fieldComputer technology
  • WIPO sectorElectrical engineering

Abstract

A method of machine learning may include receiving an API documentation including an endpoint and corresponding metadata. The method may include receiving a verified API specification including a verified endpoint. The method may include comparing the endpoint to the verified endpoint. The method may include classifying the metadata. The metadata may be classified as a positive item responsive to the endpoint being substantially the same as the verified endpoint. The metadata may be classified as a negative item responsive to the endpoint not being substantially the same as the verified endpoint. The method may include generating a training dataset, which may include the metadata classified as a positive item or a negative item. The method may include generating a metadata model using a machine learning algorithm and the training dataset. The metadata model may be configured to be used to determine whether an unknown endpoint is a positive endpoint.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.