New Web Page Standards

Most of the present limitations of the Internet spring from inadequate struc-tural information. Search engine vendors like Google have put out some suggestions for improving their performance, but we have come at this from the other end – what would it take to support a beyond-human AI problem solving engine like Dr. Eliza.

An informal survey of knowledge engines found that Dr. Eliza required just about everything that any other knowledge engine required. The following superset standard has been proposed to support all known knowledge engines. If your search engine needs something that isn’t in the following standard, then please let us know ASAP so that we can include it. Functionality not presently supported by Dr. Eliza soon will be supported.

To make this work well, you may have to chop up long web pages into sev-eral shorter pages, each one covering just one description of a condition, a treatment, or a cure, with hyperlinks to interconnect the pages.

Carefully placing the following HTML tags into your web pages will supply the information that knowledge engines need to merge your knowledge with that of everyone else following these standards, so that complex problems can be solved automatically.

‹META NAME="……..." CONTENT="…………………………………"›

Author:  Name, a comma, a space, and a professional association’s member number, to guarantee uniqueness in the world and be able to identify commonly authored knowledge, e.g. for symptom-syntax updating and merging.
Date:  This is the dating of the material in the web page and NOT the composition of the web page itself. If you are presenting a theory that is 100 years old, then the date should be 100 years earlier, e.g. September, 1908. This can be in any machine-readable but locality-independent form, e.g. 25-Sep-2008 or September 25, 2008, but NOT 25/7/08.
Content-Language  2-letter primary language, optionally followed by a dash and a 2-letter dialect (ISO 3166), e.g. “en-us”.
Code  Language 
de  German
el  Greek
en  English
es  Spanish
fr  French
hi  Hindi
it  Italian
ja  Japanese
nl  Dutch
ru  Russian
sa  Sanskrit
ur  Urdu
zh  Chinese
Rating:
Code  Movie Equivalent 
14 Years  PG-13
General  PG
Mature  R
Restricted  X
Safe For Kids  G
Condition:  Name of the cause-and-effect chain link that you are documenting. Use only letters, digits, and -
Domain-Name:  Use only letters, digits, and -
Symptom-Name:  Use only letters, digits, and -
Symptom-Syntax:  Regular expression in Dr. Eliza format.
Symptom-Search:  A symptom synonym word (or two) to throw into Google.
Symptom-Sense:  Optional “Negated” if for something that might be absent.
Symptom-Question:  A “Tell me about…” interrogatory question.
Condition-Description:  Blank field indicates to use web page contents.
Condition-Link-Type:  One of the (presently 10) types of links.
Condition-Code:  Standard code for this condition, e.g. ICD9?
Condition-Symptoms:  “/” separated list of indicative symptoms.
Condition-Counter-Indications:  “/” separated list of denying symptoms
Condition-Treatment:  Blank field indicates to use web page contents.
Condition-Cure:  Blank field indicates to use web page contents.

Note: Author, Content-Language and Rating are already in common use, but Dr. Eliza has some special needs, e.g. an association number following an author’s name.


Symptom-Syntax Control Characters
any character
any number (including zero) characters
any numeric digit
:alpha:  any alphabetic character.
.and.  fails on failure of the previous pattern, otherwise continues.
:any:  any character
:digit:  any numeric digit
:end:  not an alphabetic or numeric character, or a dash or underscore.
:enter:  a line break
:h??:  translates to the stated hexadecimal character.
:u????:  translates to the stated hexadecimal Unicode character.
:near:  presently means “following” after any number of characters.
.not.  requires that the following pattern NOT be present.
.or.  on a failure of the previous pattern, continues by testing the following pattern.
[ ]  Specifies characters or ranges of characters to match. Rules are the same as for Visual Basic’s “Like” operator.
( )  Parentheses around sub-expressions.
{ }  Braces around optional strings.
‹ ›  Around words to be recognized.

Some examples of Symptom-Syntax pattern strings: (‹amalgam›.or.‹silver›).and.(large.or.many.or.replaced) ‹margarine›.or.(‹vegetable›.and..not.(‹peanut›.or.‹olive›)) (‹glucose›.or.‹blood sugar›).and.‹1[6-9].or.‹[2-9] ‹guilt(y)›.and.(‹sin›.or.‹sins›.or.‹sinn*).or.‹punish*


Condition-Link-Type
Central  A matter directly controlled by central control system, e.g. the body's temperature.
Divine  Divine influence, without necessarily specifying the potential entity involved. This could be Satan/Devil causing a problem, or God/Allah failing to fix a problem, or any other entity with magical powers that might be involved in a problem.
Environmental  An overall controlling factor in the external environment that may then lead to an external link to the patient's physiology. This is usually something political, social, resources, temperature, power supply voltage, etc.
External  An event external to the user, that starts a cause and effect chain within the user’s direct control, potentially leading to the present symptoms. Successful identification of the head(s) of cause and effect chains is essential to finding true cures.
Hidden  A physiological change that shows few if any symptoms, but which causes other subsequent changes that may lead to the patient's present symptoms. If there is some ambiguity and this could be coded as another type of link, e.g. Intermediate, then code it as such, as these types of links are by their nature the hardest to identify.
Intermediate  A physiological change that usually shows some relatively minor symptoms, but which then causes other more serious physiological changes that may lead to the patient's present symptoms.
Proximal  Common called a diagnosis, this is the process that appears to be directly responsible for your symptoms. When your doctor says "You have ...", he has just identified your proximal link, unless of course there are several links in a self-sustaining loop, as is commonly the situation.
Symptomatic  Commonly called an idiopathic diagnosis or syndrome, a set of symptoms for which the cause may or may not be known. Symptomatic links are a statement of ignorance of the potential causal mechanisms, and so if the potential causal mechanisms are known, then these links should be reclassified, probably as Proximal.
Unknown  This is how we code things when you omit specifying a Condition-Link-Type.
Vague  A poorly understood or described process, or a collection of processes that in different manifestations could be more than one link type.

Notes:

  1. 1. Where multiple questions have been specified for the same symptom, one will be randomly selected, to avoid re-presenting the same question over and over again.
  2. 2. Your symptom-syntax may be automatically (or manually) “merged” with others to provide a good composite.
  3. 3. Note that there is NO WAY for an AI engine, or even an intelligent human without lots of domain-specific experience (as opposed to book knowledge), to compose Symptom-Syntax and Symptom-Question tags. This information is absolutely necessary for expert automated user inter-action. This impossibility dooms numerous presently ongoing “web mining” projects to failure. If you know of such a project, then please put it out of everyone’s misery by challenging it on this basis.
  4. 4. When your web page is complete, register it with Dr. Eliza by sending an email to Register@DrEliza.com with your web page’s hyperlink in the subject line.