公开(公告)号：MY181309A

公开(公告)日：2020-12-21

申请号：MYPI2015704499

申请日：2015-12-10

Applicant: MIMOS BERHAD

Inventor： KOW WENG ONN ,  DICKSON LUKOSE ,  DUC NGHIA PHAM

Abstract: A method (100) of generating knowledge cubes from multiple heterogeneous data sources comprises the steps of analyzing all the data sources to identify potential data cubes and parameters (S202), defining (S208) cubes through the use of natural language description (S201) to obtain query results (S503) using the identified data cubes and parameters (S202), aggregating (S510) the query results upon normalization of the query results with parameters, and populating (S508) the data cubes from the heterogeneous data sources by aggregating (S510) and harmonizing the measurements of the data source. The populated cube is stored (S509) in the harmonized results database (608) for sharing and reusability.

公开(公告)号：MY188280A

公开(公告)日：2021-11-25

申请号：MYPI2015704111

申请日：2015-11-13

Applicant: MIMOS BERHAD

Inventor： ABDUL KALIQUE SHAIKH ,  ARUN ANAND SADANANDAN ,  DUC NGHIA PHAM ,  DICKSON LUKOSE

IPC: G06N5/00 ,  G06F17/00

Abstract: The present invention identifies implicit causal relationships by analysing relationships between entities from medical reports such as medical history and patient personal profile. The system (100) of the present invention for indentifying implicit causal relationships from medical reports using network analysis comprises knowledge base; and at least five modules. The at least five modules comprising a term extraction module for extracting a set of terms with entity type identification from document set using entity recognition; a semantic parsing module for performing semantic parsing to identify relevant matching concepts from ontological resources using semantic matching; a network generation module for identifying co-occurrence relations and building network based on co-occurrence relations; an importance analysis module for determining importance measure of terms using social network analysis functions; and a causal relations extraction module for executing entity type analysis to identify implicit cause-effect relations. The present invention discovers hidden associations which are not apparent among terms and concepts of medical reports through generation of semantic network of concepts from medical reports and to perform importance analysis and entity type analysis.