Combining data driven models for classifying data
AssigneeBASF SE
InventorsArunav Mishra, Juergen Mueller, Lalita Shaki Uribe Ordonez
Life of this patent
Estimating the term…
Abstract
The present invention relates to classifying data (24). A first data driven model (50) is trained based on labeled historic data (44). A second data driven model (60), comprises a set of rules (42). Data (24) to be classified is obtained at the first data driven model (50) and the second data driven model (60). A first classification (52) is determined for the data (24) by the first data driven model (50) and a second classification (62) is determined for the data (24) by the second data driven model (60). A result signal (80) is provided based on the classifications (52, 62).
Claims15
System (10) for classifying data (24) comprising a communication interface (14), and a processor (18), the system (10) configured for:
- providing a first data driven model (50), wherein the first data driven model (50) is trained based on labeled historic data (44),
- providing a second data driven model (60), wherein the second data driven model comprises a set of rules (42),
- obtaining data (24) at the first data driven model (50) and the second data driven model (60) via the communication interface (14),
- determining a first classification (52) for the data (24) by the first data driven model (50),
- determining a second classification (62) for the data (24) by the second data driven model (60), and
- providing a result signal (80) based on the classifications (52, 62).
Computer implemented method (800) for classifying data (24), comprising the steps:
- providing a first data driven model (50), wherein the first data driven model (52) is trained based on labeled historic data (44),
- providing a second data driven model (60), wherein the second data driven model (60) comprises a set of rules (42),
- obtaining data (24) at the first data driven model (50) and at the second data driven model (60),
- determining a first classification (52) for the data (24) by the first data driven model (50),
- determining a second classification (62) for the data (24) by the second data driven model (60), and
- providing a result signal (80) based on the classifications (52, 62).
Dependencies are inferred from the claim text.