Why the Semantic Web, Twine, and the like are not enough!

Humans have developed language allowing for abstraction of perception, or, in other words, semantic memory. The invention of script allowed for semantic knowledge explication via human artifacts. Semantic communication thus could become asynchronous and spatially extended, extending beyond the immediate reach of voice. Since the inventions of voice recording and voice transmission, the same can be maintained for spoken language. However, the hitherto central means of semantic communication has not changed at all: natural language expressions. Due to limitations of parallelism of motor actions and, closely related, consciousness, the preferred way of semantic communication still is a serialization of natural language expressions in the form of documents or speeches. The resulting (flattened) organization of semantic information has to be reconstructed in its original form and woven in the (hierarchic-conceptual / networked-associative / dynamic) semantic memory of the receiving party. This, as a matter-of-fact, considerably restricts human semantic communication, by, in a way, enabling sender-associativity only, not naturally receiver-associativity. Not even modern time's Search technology in conjunction with the Internet is able to tackle this fundamental weakness because (at best receiver-associative) search results use to point to de-associated serialized natural language expressions (documents, video/speech recordings). The Semantic Web movement led by Tim Berners Lee, which seems to not be quite aware of the fundamental dimension of this problem, takes a false step into the right direction by translating documented semantic knowledge into networked semantic expressions thus trying to create a general semantic knowledge Web. The fallacy of this approach is the neglect of other central aspects of semantic communication than the structural alone. Firstly, there is the dynamic nature of semantic knowledge / memory. It changes over time by knowledge / memory adaptation/assimilation, while artifacts tend to stay the same. Thus a semantic-structural gap develops over time necessitating repeated communication by devaluation of knowledge artifacts once communicated. Furthermore, the fundamentally contextual-structural nature of semantic information lets one become doubtful about the idea of generating a general, de-contextualized Semantic Knowledge Web (that ideally would somehow integrate all world knowledge, as it seems) which is rooted in the symbolic AI movement of earlier decades and is thus prone to meet the same criticism as faced by GOFAI in the past.

In ‘Artificial Memory’, I describe/explore a new approach to semantic communication and information management by applying fundamental insights of memory research onto information technology in order to allow for a paradigm shift in human semantic communication via a new cultural technique (artificially memorizing). An artificial memory system is to duly reflect an individual's dynamic semantic knowledge/memory. It ought to be threefold: reflective (reflecting consciousness), extending (by associating to assertions part of biological memory), and expanding (by associating/interconnecting one's own semantic structures to other semantic structures). Important questions have to be tackled: In how far is extending semantic mind by artificial memory systems possible? Can it form a transcranial system? What about emotional connotations? How can automaticity of thoughts, feelings, and actions be reflected in artificial memories? Etc.

Though many different fields of research (cognitive science, information science, psychology, Human-Computer-Interaction, philosophy) have advised ‘Artificial Memory’, I think that - overall - the fresh perspective on semantic communication opens a new field of research in itself.