Evolving knowledge through negotiation

Semantic web information is at the extremities of long pipelines held by human beings. They are at the origin of information and they will consume it either explicitly because the information will be delivered to them in a readable way, or implicitly because the computer processes consuming this information will affect them. Computers are particularly capable of dealing with information the way it is provided to them. However, people may assign to the information they provide a narrower meaning than semantic technologies may consider. This is typically what happens when people do not think their assertions as ambiguous. Model theory, used to provide semantics to the information on the semantic web, is particularly apt at preserving ambiguity and delivering it to the other side of the pipeline. Indeed, it preserves as much interpretations as possible. This quality for reasoning efficiency, becomes a deficiency for accurate communication and meaning preservation. Overcoming it may require either interactive feedback or preservation of the source context. Work from social science and humanities may help solving this particular problem.

💡 Research Summary

The paper positions semantic‑web information at the far ends of a long “pipeline” that connects human beings and computational systems. Humans are both the originators and consumers of this information: they may read it directly when it is presented in a human‑readable form, or they may be affected indirectly when software that consumes the data influences their environment or decisions. Computers, by contrast, process information exactly as it is supplied, without any built‑in assumptions about the author’s intended narrow meaning.

A central tension highlighted by the authors is that people often treat their own statements as unambiguous, believing that the meaning they intend is the only one that matters. Model theory, which underlies most formal semantics on the semantic web, deliberately preserves every logically possible interpretation of a statement. This design choice maximises reasoning efficiency and logical completeness, but it also creates a communication gap: the receiver (human or machine) is presented with a set of potential meanings far broader than the author’s intended one, leading to ambiguity, misinterpretation, and loss of precise meaning.

To bridge this gap the authors propose two complementary strategies. The first is interactive feedback, a negotiation loop in which the human and the system iteratively refine the meaning of a statement. In practice this could be realized through question‑answer dialogues, explainable‑AI (XAI) explanations, or adaptive user interfaces that ask for clarification when multiple interpretations are detected. By allowing the user to confirm, reject, or modify the system’s provisional reading, the pipeline gradually narrows the set of admissible models until the author’s intended meaning is isolated.

The second strategy is preservation of source context. Instead of transmitting raw triples or axioms alone, the system also conveys metadata about the author’s background, the discourse situation, the lexical choices, and any pragmatic cues that shaped the utterance. Ontological annotations, provenance records, and contextual embeddings can serve as carriers of this information. When a downstream agent receives both the formal statement and its surrounding context, it can apply a more informed disambiguation process that respects the original intent.

The paper argues that insights from the social sciences and humanities are essential for designing these negotiation mechanisms. Pragmatics in linguistics studies how speakers and listeners negotiate meaning in real time, offering models for turn‑taking, clarification requests, and repair strategies. Sociology’s theories of social construction emphasize that shared meanings emerge through collective negotiation rather than being fixed a priori. By importing these concepts into the technical architecture of the semantic web, we can align the formal, model‑theoretic preservation of multiple interpretations with the human need for precise, context‑sensitive communication.

In summary, the authors contend that knowledge on the semantic web should be viewed as evolving through negotiation rather than as a static transmission of facts. Interactive feedback loops and rich contextual metadata together enable a dynamic refinement of meaning, turning the “pipeline” from a one‑way conduit into a bidirectional dialogue. This approach not only improves the fidelity of human‑computer communication but also advances the broader goal of making the semantic web a truly interoperable, socially aware information infrastructure.