公开(公告)号：US20240330329A1

公开(公告)日：2024-10-03

申请号：US18127963

申请日：2023-03-29

申请人： Kyndryl, Inc.

发明人： Archana Dixit ,  Amol Bhaskar Mahamuni ,  Kiran Kumar K ,  Mahesh Sondkar

IPC分类号： G06F16/29 ,  G06F16/2457 ,  G06F16/28

CPC分类号： G06F16/29 ,  G06F16/24578 ,  G06F16/285

摘要： A computer-implemented method, in accordance with one embodiment, includes collecting data from a plurality of predefined functional areas in a particular geographical area in which a plurality of physical objects are located, at least some of the data relating to vandalism. Wrangling is performed on the data to correlate the data from the functional areas with respective ones of the physical objects. The wrangled data corresponding to each of the physical objects is transformed into features corresponding to a the functional areas. For each of the physical objects, the features are processed according to predefined rules for generating a rank for each of the functional areas. Dynamic weights are computed for at least some of the features to account for data drift by comparing historical data with latest data. The ranks and dynamic weights are transformed into a vandalism score for each of the physical objects, which is output.

公开(公告)号：US12111855B1

公开(公告)日：2024-10-08

申请号：US18127963

申请日：2023-03-29

申请人： Kyndryl, Inc.

发明人： Archana Dixit ,  Amol Bhaskar Mahamuni ,  Kiran Kumar K ,  Mahesh Sondkar

IPC分类号： G06F16/00 ,  G06F16/2457 ,  G06F16/28 ,  G06F16/29

CPC分类号： G06F16/29 ,  G06F16/24578 ,  G06F16/285

摘要： A computer-implemented method, in accordance with one embodiment, includes collecting data from a plurality of predefined functional areas in a particular geographical area in which a plurality of physical objects are located, at least some of the data relating to vandalism. Wrangling is performed on the data to correlate the data from the functional areas with respective ones of the physical objects. The wrangled data corresponding to each of the physical objects is transformed into features corresponding to a the functional areas. For each of the physical objects, the features are processed according to predefined rules for generating a rank for each of the functional areas. Dynamic weights are computed for at least some of the features to account for data drift by comparing historical data with latest data. The ranks and dynamic weights are transformed into a vandalism score for each of the physical objects, which is output.