Collaborative Discussion 1: Week 1

Initial Post

by Abdulhakim Bashir - Monday, 5 May 2025, 11:53 AM

The assertion that knowledge representation (KR) emerged only with computing technology overlooks its rich historical roots. So, I respectfully disagree with the assertion. While formal KR systems in AI are indeed recent developments, humans have been creating external representations of knowledge for millennia and historical evidence indicates that KR has deep roots in human intellectual traditions. For instance, Weststeijn (2011) discusses how early modern Dutch scholars engaged with pictographic systems, such as hieroglyphs, to develop universal characters aimed at representing knowledge visually and universally. Examining Weststeijn’s (2011) analysis of early modern pictography reveals how scholars sought universal visual languages to represent knowledge across cultural boundaries, demonstrating that KR principles predate computing and further emphasizes a longstanding human pursuit to codify and communicate knowledge systematically.

In the field of artificial intelligence (AI), knowledge representation (KR) is intrinsically linked to reasoning. Brachman and Levesque (2004) emphasize that KR provides the structures necessary for reasoning processes, enabling systems to infer new information and make decisions. Without reasoning, KR would merely be static data structures lacking the dynamic capability to derive conclusions or adapt to new information. Davis, Shrobe, and Szolovits (1993) further elaborate that KR serves multiple roles, including facilitating reasoning, guiding inference, and supporting efficient computation.

Therefore, reasoning is not just related to KR; it is essential for its functionality and utility, and the contemporary understanding of KR in computing extends ancient practices rather than inventing them. What distinguishes modern KR is its formalization and algorithmic implementation, not its conceptual foundation.

References:

Brachman, R. J., & Levesque, H. J. (2004). Knowledge Representation and Reasoning. Morgan Kaufmann.

Davis, R., Shrobe, H., & Szolovits, P. (1993). What is a Knowledge Representation? AI Magazine, 14(1), 17–33.

Weststeijn, T. (2011). From hieroglyphs to universal characters: Pictography in the early modern Netherlands. Netherlands Yearbook for History of Art, 61(1), 238–281.

Peer Response

by Nikolaos Archontas - Thursday, 8 May 2025, 3:24 PM
In reply to Abdulhakim Bashir

I fully agree that knowledge representation (KR) is not a recent innovation but rather an ancient human practice rooted in the necessity to encode, preserve, and transmit knowledge. Long before computational KR, early civilizations developed sophisticated symbolic systems—from Paleolithic cave paintings (~40,000 BCE) to Egyptian hieroglyphics (~3300 BCE)—that served as durable, interpretable repositories of cultural, religious, and administrative knowledge (Weststeijn, 2011). These systems underscore humanity’s enduring drive to externalise thought, a pursuit later refined through alphabets, formal logic (i.e. Aristotle’s syllogisms, Leibniz’s characteristica universalis), and mathematical notation (Brachman & Levesque, 2004). Such historical precedents confirm that KR’s conceptual foundations predate computers by millennia and modern AI merely formalises these principles algorithmically.

However, while KR can exist as a standalone framework for organising and retrieving data (structured databases or ontologies) its full potential is realised only when paired with reasoning. As Brachman and Levesque (2004) argue, KR structures (i.e. semantic networks, frames) provide the what (facts, rules, relationships), while reasoning enables the how (inference, problem-solving). For instance, a medical ontology might encode disease-symptom relationships, but without reasoning (i.e. deductive logic or probabilistic inference), it cannot diagnose new cases or resolve contradictions (Davis et al., 1993). This aligns with Sowa’s (2000) observation that reasoning transforms static KR into dynamic intelligence, generating actionable insights (Sowa, 2000).

Reasoning may enhance KR by enabling inference and problem-solving, but KR alone still has value. Systems like databases and knowledge graphs show that structured KR works well for storing and finding information, even without advanced reasoning (Sowa, 2000). The real challenge for AI today is combining these modern tools with principles of clear representation to create systems that are both powerful and understandable.

References

Brachman, R. J., & Levesque, H. J. (2004). Knowledge Representation and Reasoning. Morgan Kaufmann.

Davis, R., Shrobe, H., & Szolovits, P. (1993). What is a Knowledge Representation? AI Magazine, 14(1), 17–33.

Sowa, J. F. (2000). Knowledge Representation: Logical, Philosophical, and Computational Foundations. Brooks/Cole.

Weststeijn, T. (2011). The Visible World: Dutch Visual Culture and the Rise of Early Modern Science. Amsterdam University Press.

Peer Response

by Abdulrahman Alhashmi - Monday, 12 May 2025, 1:02 PM
In reply to Abdulhakim Bashir

I support your stance on the question of whether Knowledge Representation (KR) emerged with computing technology. Indeed, the former has strong historic roots. You backing-up this information with viable peer-reviewed sources such as Weststeijn (2011) who discusses how Dutch scholars who explored visual language in pictography in the exploration of language. In line with the above analysis, it is evident that the KR principles outdate computing technology. Through KR, the ability to decode and communicate knowledge has been a possibility. To ascertain that knowledge is communicable in KR in AI I think is the fundamental goal of systematic representation even through visualization.

Secondly, in your analysis of the second question, I agree with the positive linkage of knowledge representation and reasoning. I particularly like the sources used in your analysis-Brachman and Levesque (2004) and Davis aet al. (1993). In line with their respective analysis, “KR provides the structures necessary for reasoning processes, enabling systems to infer new information and make decisions.” The absence of reasoning KR does not make as much sense. The distinction you have made between modern forms of computing and historic endeavor shows the evolutionary nature of KR and AI.

I support the conclusion of your analysis; indeed, reasoning is critical for, functionality and utility, and the contemporary understanding of KR in computing. KR is a continuation of ancient practices through its distinct conceptual foundation, as compared to new inventions.

Reference:

Brachman, R. J., & Levesque, H. J. (2004). Knowledge representation and reasoning. Morgan Kaufmann.

Weststeijn, T. (2011). Picturing knowledge in the early modern Netherlands. Journal of the History of Ideas, 72(4), 547–568.

Davis, R., Shrobe, H. E., & Szolovits, P. (1993). What is a knowledge representation? AI Magazine, 14(1), 17–33.

← Back to Unit 1

KRR Module Home