Are our knowledge graphs trustworthy?
0 likes602 views
The document discusses the trustworthiness of knowledge graphs. Knowledge graphs are machine-readable data organized for general purpose use in applications like search engines, recommender systems, and virtual assistants. While knowledge graphs are used alongside machine learning in AI systems as a source of domain knowledge, it is unclear if they can be trusted. To determine if knowledge graphs are trustworthy, we must consider both the process of knowledge engineering used to create them as well as the knowledge graph as an end product. Factors like understanding the data's provenance, auditing it for bias, transparency in its creation and use are important for establishing trust.
Related topics:1 of 12
Download to read offline
More Related Content
What's hot (20)
Ad
Similar to Are our knowledge graphs trustworthy? (20)
Ad
More from Elena Simperl (20)
Are our knowledge graphs trustworthy?
- 1. Are our knowledge graphs trustworthy? International workshop on knowledge science March 2022, online Elena Simperl King’s College London @esimperl
- 2. Knowledge graphs are machine- readable data organised for general-purpose use
- 3. Applications in search, recommenders, virtual assistants, enterprise data integration etc. Used in AI systems alongside machine learning as source of domain knowledge, transfer learning, explanations
- 5. What do we mean by trust? Process (knowledge engineering) vs outcome (knowledge graph as a resource) Frameworks, methods, guidance on trustworthy AI, data quality, data interoperability standards
- 6. We know how good the data is
- 7. We know where the data comes from
- 8. We audit our data to make it less biased
- 9. We know how the data was created
- 10. We know who created the data
- 11. We know how the data is used
- 12. Empirical knowledge engineering Knowledge graphs and language models Regulatory compliance This Photo by Unknown Author is licensed under CC BY-NC-ND