Wiki science/Semantic Prosthetic

Sefmantic Networks and Personal Experience

Human conscious experience is normally dominated by memory. New sensory experiences are processed in ways that are heavily influenced by past experiences. When sensory stimuli are reduced (sensory deprivation), a human brain is able to continue generating chains of thoughts by making use of existing memories. The ways in which one thought leads to another thought reflects a semantic network that exists within the brain. The behavior of such a semantic network can be described as a content addressable memory system. A new sensory pattern or a thought will "resonate" with existing contents of the memory system leading to activation of relevant memories and the generation of adaptive thoughts and behavioral responses. The semantic memory network is modified by on-going thoughts resulting in new memories. With time, human semantic memory comes to contain an increasing complex representation of the world.

Theory-based Semantics, Language of the Brain

Theory-based semantics states that the binding element of human thought is "theory," which constrains the meaning of concepts, ideas and thought patterns stored within the brain, according to their associative relationships to other concepts and ideas. As a result, concept, idea and thought pattern meanings are declarative and language independent, experienced as a "language of thought."

According to Dr. Richard L. Ballard, theory represents more than 85% of knowledge, withe the who, what, when, where and how much facts of situations and circumstances representing only 15%. Theory is “a priori” (known before the fact). It is the element that gives meaning to concepts, ideas and thought patterns, and the conditional reasoning power required to answer our "How," "Why" and "What if" questions. Learned through enculturation, education, life experience and deep analytical thinking, theory shapes our behavior and the way we understand our world. Well-justified theories such as those proven most successful by science, engineering, business and law are most valuable. Theory is predictive and considers all possibilities. Once it is learned it is used for decades, centuries and millenniums. Most of the core theories that shape our communial social behavior, for example, were conceived 10,000 to 24,000 years ago, passed down through the generations. Much of our financial theories, such as "buy low, sell high," or "the principles of interest," were conceived and put into use by our ancestors millenniums ago. Circumstances and situations may change rapidly, but the underlying theory that gives them meaning, do not.

First proposed by Richard L. Ballard, Ph.D. between 1987-1993, theory-based semantics is the governing principle of theory-based semantic systems which allow semantic software-based machines to reason with the same theories that humans use. Reference Source: "Fundamental Definitions in Knowledge Science & Engineering" by Dr. Richard L. Ballard, 12/2004. Physical Theory of Knowledge and Computation (2006).

Human Language as a Semantic Prosthetic Device

We all live in the same world and we all share a common genetic heritage which determines the basic structure and function of the brain. Humans have evolved in a social environment where there is a high level of reward for knowing what other people are thinking. It has been suggested that this need to know what others know is the major reason for our unique human cognitive strengths. If so, it may be that human languages evolve so as to best serve as tools for allowing people to come to know what other people are thinking. We can think of human language as a cultural artifact, a prosthetic device or a tool that assists humans to survive by enhancing their ability to come to know what other people know. We can describe human language as the original semantic prosthetic.

Wiki as a prosthetic device

Written language is a relatively new invention. Books and other written documents have become useful aids for the cultural transmission of knowledge and they certainly qualify as useful semantic prosthetics. However, traditional books and printed documents have severe limitations as semantic prosthetics: they are limited in perspective, expensive, rare, hard to access, and they easily become out-dated. Electronic forms of communication and knowledge exchange can avoid some of the limitations of printed books. When distributed by the world wide web, electronic documents such as HTML pages can become cheap, easy to copy, and easy to access. A single individual can keep conventional HTML pages up to date, but a quick look at the World Wide Web (WWW) reveals a common problem: too many HTML pages that are abandoned by their single authors and which quickly become irrelevant fossils.

Wiki HTML pages can escape the limitations of narrow perspective imposed by a limited number (usually just one) of authors and also make it easier to keep pages up to date by sharing the work load involved in the update process. While conventional HTML pages can be cheap, another all too common feature of the WWW is the sale of information. Open and free modes of information exchange (like wikis using a free license) can place reasonable limits on the cost of obtaining information. It remains to be seen just how dramatically human communication will be affected by new semantic prostheses like wikis. Empirical observations will be required as part of the scientific study of Wikis as semantic prostheses.

Semantic prosthetic theory and the future of Wiki

Can we make fruitful guesses about the future of wikis as semantic prosthetic devices? I suggest that a theoretical foundation for the science of wiki can be established by formalizing the idea of a semantic network. We can start by thinking about the set of all things that people have known, now know, and will know in the future. We can conceptualize this set of human knowledge as a high-dimensional space, but project all of the high-dimensional structure into a simple 2-dimensional map for "back of the envelop" evaluation of basic ideas.

One important 2-dimensional projection results from mapping each person's total knowledge as a point on a 2D "everything I know" map (or, for short, "knowledge map"). We imagine that at any given time, each individual is located at some point (A) on this map. A prosthetic device like a wiki can speed the rate at which a person moves across the map towards a new desired state of knowledge (B). In building our wikis, we can seek ways to maximize their ability to increase the speed at which people can move around on the "everything I know" map. The formation of new useful semantic memory (movement across the "everything I know" map) depends on a fruitful matching (path 2 in the diagram) between what a person already knows and the format of the information provided by the semantic prosthetic device that is guiding that person to a new state of knowledge (a new location on the knowledge map). For each person the topography of the knowledge map might be different and for most efficient movement to a new state of knowledge there is an optimal path. Without good matching between the learner and the topic to be learned, the learning process is more difficult (path 1 in the diagram).

How can a wiki evolve so as to be able to provide optimal paths? I think there are two basic ways to promote the utility of a wiki as a semantic prosthetic device. First, the people who make the wiki pages have to be thinking about how to most efficiently move a person from point A to point B. This requires an understanding of what it means for a person to be at point A or point B and an understanding of the cognitive landscape around this region of the knowledge map. Second, the people making the wiki pages have to be able to abandon the idea of "one page fits all". Different people are going to arrive at point B from different starting points. Diversity, a multiplicity of approaches to point B need to be made available. This is the positive approach and algorithm for wiki success.

Pulling our ladders up behind us

I think the major problem for optimization of a wiki is that once people reach a certain knowledge state, they tend to forget what it was like before they got there. This problem can be addressed by collaboration between people who are at point A and point B on a knowledge map. Users of a wiki who are using the wiki to get from A to B have to be able to provide feedback to those people who are already at B. By way of this feedback learners can make known the places on a wiki page where they fail to make rapid progress towards their learning goal. At the very least, there needs to be a way for people to mark places on a wiki page where they get confused and need more help. Theoretically, such feedback could take place on "talk" pages that are associated with each content page. However, current wiki talk pages tend to be dominated by arguments between people who are already at point B.

Actual users (those who are learning from the content of the page) need a simple and quick way to mark difficult locations in a page. The standard for ease of use is "point and click." I propose that wiki engineers should develop a system by which wiki readers can be provided with a small pallet of "feedback markups" where they can simply click on an icon and "stick" it on the page where they have a problem or want to make a comment. Ideally, each page could have a transparent "markup" overlay that would display the user feedback icons and could be used to quickly visually identify problem locations on a page.

The "stop", "thumbs up" and "thumbs down" icons would just be dragged onto the page and dropped. When the "add a comment" icon is dropped, a comment window would open allowing the user to add a comment for that particular location in the page. An example of a wiki page with this user markup overlay is shown below.