Semantic Web Technologies and Bias in Artificial Intelligence: A Systematic Literature Review

Tracking #: 3041-4255

Authors: 
Paula Reyero Lobo
Enrico Daga
Harith Alani1
Miriam Fernandez1

Responsible editor: 
Dagmar Gromann

Submission type: 
Survey Article
Abstract: 
Bias in artificial intelligence (AI) is a critical and timely issue due to its sociological, economic and legal impact, as decisions made by biased algorithms could lead to unfair treatment of specific individuals or groups. Multiple surveys have emerged to provide a multidisciplinary view of bias or to review bias in specific areas such as social sciences, business research, criminal justice, or data mining. Given the ability of Semantic Web (SW) technologies to support multiple AI systems, we review the extent to which semantics can be a "tool" to address bias in different algorithmic scenarios. We provide an in-depth categorisation and analysis of bias assessment, representation, and mitigation approaches that use SW technologies. We discuss their potential in dealing with issues such as representing disparities of specific demographics or reducing data drifts, sparsity, and missing values. We find research works on AI bias that apply semantics mainly in information retrieval, recommendation and natural language processing applications and argue through multiple use cases that semantics can help deal with technical, sociological, and psychological challenges.
Full PDF Version: 
Tags: 
Reviewed

Decision/Status: 
Accept

Solicited Reviews:
Click to Expand/Collapse
Review #1
By Konstantinos Kotis submitted on 07/Mar/2022
Suggestion:
Minor Revision
Review Comment:

In this revised version, authors have answered in a sufficient way to all comments provided by the reviewers. The paper has been significantly improved, although some minor issues (in language) still remain (please have a careful check line-by-line). Some issues also with the visibility of figures must be resolved.

I am still confident that, although there are several works out there presenting results related to the AI bias domain, this systematic review paper is an excellent source for new (and experienced) researchers to review the current trends and challenges of this research area, especially in relation to the area of Semantic Technologies.

Review #2
By Dagmar Gromann submitted on 11/May/2022
Suggestion:
Accept
Review Comment:

Thank you very much for having addressed all of the reviewers' comments in great detail. I believe the resulting revision to be largely improved and very interesting for the AI and Semantic Web community, as it provides a very well structured and argued overview of their interaction to counteract bias.

The language and style of presentation is largely improved, apart from the very minor comments below. Two last aspects that should be considered in the final version are: 1) the (L)LOD cloud contains ontologies, lexicons, thesauri - this distinction between KGs, lexicons, etc. and LOD does not make a lot of sense - please add that all of these resources can and are represented as LOD. 2) the categories in Table 3 are highly data-centric. Could you please add a reason for that and why these are then not re-used in the detailed presentation of surveyed appraoches?

Minor comments:
In general: , i.e., and , e.g. are separated by commas from the main sentence and not brackets
p. 1, 38 platforms such as => platforms, such as
p. 2, 23 In this study => In this article/survey
p. 2, 44 areas under semantic research => not sure what "under" means here
p. 4, 27 encode as code with \texttt?
p. 14, 3 possible relations between them and the possible relations between them => copy-paste error?
p. 18, 37 bias [65] bias