Contextual knowledge graph approach to bias-reduced decision support systems

Year:	2024
Title:	Contextual knowledge graph approach to bias-reduced decision support systems
Abstract:	Proposes a contextual KG approach to capture relationships between task/features/context, identifying bias in AI/ML model datasets. Uses debiased datasets for fairer decision-making.
Objectives:	To identify bias in AI/ML datasets and support fairer decision-making.
Methodology:	Contextual KG approach; proposed three bias assessment metrics (label, sampling, timeliness bias).
Algorithm Used:	N/A (Methodology, leverages ML models)
Top Model:	Contextual Knowledge Graph Approach
Accuracy:	More effective in supporting fairer decision-making than existing methods (Experimental results)
Advantages:	Combines contextual knowledge for fairer decisions; defines specific bias metrics (label, sampling, timeliness bias).
Drawbacks:	N/A explicitly mentioned.
Source:	https://www.tandfonline.com/doi/full/10.1080/12460125.2024.2349436#abstract Edit

Criteria (7)

Name	Description	Definitions
Objectivity (Bias	Proposes a contextual KG approach to capture relationships between task/features/context, identifying bias in AI/ML model datasets. Uses debiased datasets for …	Proposes a contextual KG approach to capture relationships between task/features/context, identifying bias in AI/ML model …
Objectivity (Bias)	Quality dimension from Contextual knowledge graph approach to bias-reduced decision support systems	Quality dimension from Contextual knowledge graph approach to bias-reduced decision support systems (2024)
Timeliness	Proposes a contextual KG approach to capture relationships between task/features/context, identifying bias in AI/ML model datasets. Uses debiased datasets for …	Quality dimension from Contextual knowledge graph approach to bias-reduced decision support systems (2024) Proposes a contextual KG approach to capture relationships between task/features/context, identifying bias in AI/ML model …
Accuracy	Proposes a contextual KG approach to capture relationships between task/features/context, identifying bias in AI/ML model datasets. Uses debiased datasets for …	Quality dimension from Contextual knowledge graph approach to bias-reduced decision support systems (2024) Proposes a contextual KG approach to capture relationships between task/features/context, identifying bias in AI/ML model …
Relevancy	Proposes a contextual KG approach to capture relationships between task/features/context, identifying bias in AI/ML model datasets. Uses debiased datasets for …	Quality dimension from Contextual knowledge graph approach to bias-reduced decision support systems (2024) Proposes a contextual KG approach to capture relationships between task/features/context, identifying bias in AI/ML model …
Value added	Proposes a contextual KG approach to capture relationships between task/features/context, identifying bias in AI/ML model datasets. Uses debiased datasets for …	Quality dimension from Contextual knowledge graph approach to bias-reduced decision support systems (2024) Proposes a contextual KG approach to capture relationships between task/features/context, identifying bias in AI/ML model …
Fairness	Proposes a contextual KG approach to capture relationships between task/features/context, identifying bias in AI/ML model datasets. Uses debiased datasets for …	Quality dimension from Contextual knowledge graph approach to bias-reduced decision support systems (2024) Proposes a contextual KG approach to capture relationships between task/features/context, identifying bias in AI/ML model …