PubWanDataDictionary

• PubWanDataEntryLayer
• PubWanNormalizationLayer
• PubWanAbstractionLayer

I added SandBox, the rest appeared magically. See the latest, http://www.xanthus.net/phpAdmin user PubWan pswd StarLee, select a table, Browse the data, create. -- JimScarver

Database PubWan

# phpMyAdmin MySQL-Dump
# version 2.2.0
# http://phpwizard.net/phpMyAdmin/
# http://phpmyadmin.sourceforge.net/ (download page)
#
# Host: localhost
# Generation Time: December 29, 2002, 11:13 am
# Server version: 3.23.40
# PHP Version: 4.1.2
# Database : `PubWan`
# --------------------------------------------------------

#
# Table structure for table `SandBox`
#

CREATE TABLE SandBox (
  who varchar(80) default NULL,
  what text,
  when timestamp(14) NOT NULL,
  where text,
  why text,
  how text,
  quantity int(11) NOT NULL default '1'
) TYPEMyISAM COMMENT'This is for playing with SQL';
# --------------------------------------------------------

#
# Table structure for table `adjective`
#

CREATE TABLE adjective (
  key bigint(20) default NULL,
  adjective text
) TYPE=MyISAM;
# --------------------------------------------------------

#
# Table structure for table `brand`
#

CREATE TABLE brand (
  key bigint(20) default NULL,
  brand text
) TYPE=MyISAM;
# --------------------------------------------------------

#
# Table structure for table `noun`
#

CREATE TABLE noun (
  key bigint(20) default NULL,
  noun text
) TYPE=MyISAM;
# --------------------------------------------------------

#
# Table structure for table `package`
#

CREATE TABLE package (
  key bigint(20) default NULL,
  system tinyint(4) default NULL,
  upc1 bigint(20) default NULL,
  upc2 bigint(20) default NULL,
  check tinyint(4) default NULL,
  brand bigint(20) default NULL,
  noun bigint(20) default NULL,
  net_mass float default NULL
) TYPE=MyISAM;
# --------------------------------------------------------

#
# Table structure for table `package_adjective`
#

CREATE TABLE package_adjective (
  key bigint(20) default NULL,
  package bigint(20) default NULL,
  adjective bigint(20) default NULL
) TYPE=MyISAM;

</verbatim>


I| admit to this outrage.  It was done in haste.

I made the relationship between packages and adjectives many-to-many because I see "canned creamstyle yellow corn (maize)" as a subset of "creamstyle yellow corn", "canned yellow corn" and "canned creamstyle corn".  Likewise "creamstyle yellow corn" is a subset of "creamstyle corn" and "yellow corn".  There is also a need to encode knowledge like "corn" being a subset of "vegetable" (in at least some cultural senses) and "cereal" (in a more scientific sense).

Two brand names in one cell suggests that also should be in a many-to-many relationship with packages.  Ideally we could catalog subsidiary relationships between brands, and mappings between brands and business entities.  This, of course would be every bit as tedioous as the original PubWan objective of creating an informational lens to focus on the supply side of the consumer marketplace.  We wouldn't want to get bogged down in "database of everything" ambitions.

Tabulating brands separately from (other) adjectives may be redundant, especially with the inclusion of ® symbols.

At this stage of development, major restructuring should be done as freely as data addition.  Hopefully we can maintain some kind of audit trail, and an open conversation on experimentation, both before and after the fact.  For this PubwanChat as suggested by StarPilot would be indispensable.

I had some difficulty with the user interface, as well as the SQL dialect (I'm more familiaqr with Oracle®).  I couldn't get in any requests for non nullability, primary key designation or indexing without error mensajes.  Rationalizing the layout, and adding population would seem a higher priority for now anyway.
  'Hint: Be sure to select the right datatypes type first and insure that any  data in the table row is consistant with the options you are setting.'

MOT, http://mot.sourceforge.net/ seems like it might be good for PubWan, I downloaded it, but it would not initialize properly.  If someone else thinks this would be helpfull and anyone wants to fight with it, i'll give them the password..... -- JimScarver

Some idle speculations on pubwan data model.

In general, I| would recommend using SI units for all weights and measures.  This is internally in pubwan base tables.  User interfaces must of course be adaptible to any system of units.  It would be awkward (in consumer math) to measure volume in cubic meters or time in seconds.

pubwan data tables

'package:key,upc,desc,netmass,netvol'

Basically, a package is a gproduct equivalence class.  Industry uses the UPC as a unique indentifier of packages most of the time.  Pubwan should make note of UPC's, but should assign its own unique keys.  One challenge in implementing pubwan will be coordination of these between nodes of a distrubuted database.

'desc:key,desc,gendesc'

The desc field of a package object should refer to a very specific descriptor of the package.  The desc table is a sort of category tree.

'catch:key,time,package,grossmass,pkgmass'

This is for catch weighting.  Net_mass is (nominal net mass) is the same for all instances of the packages, whereas these will be multiple empirical observations.  For products sold by volume this will also give information about density, information not usually found on packages and that  can be useful for recipes and the like.


'sale:key,time,volunteer,vendor,package,price'

'business:key,name,addr,loc,parent'

'addr:key,no,box,street,city,state,zip,country'

'loc:key,lat,long'

One of the most tedious (and certainly controversial) of pubwan development activities.  Manual data entry of receipts will probably be the usual method for populating this table early on.  Gathering information without buying will most certainly be challenged, probably on intellectual property grounds.

'shoplist:key,volunteer,volunteerp,desc,budget,time'

'volunteer:key,handle,handlep,name,namep,addr,addrp,loc,locp'

These objects contain personal information about pubwan volunteers.  Following common lisp nomenclature convention, the suffix "p" indicates boolean data type.  Fields tagged "false" will not be made publically available by pubwan.

'spring:key,volunteer,desc1,desc2,accel'

This database is for spatial modeling projects, probably not of interest to pubwan in its infancy due to needs for large datasets and sophisticated algorithms.  Accel or d2r/dt2 would be negative for attraction and positive for repulsion.

'packagedims:key,package,length,width,height_

'packagetemp:key,desc,temp_

One spatial modeling project I hope will eventually be feasible is a "virtual store".  Picture a 3D CAD model of packages on shelves with a visual walkthrough.  The operative word is definitely virtual.  In the virtual world, things are possible that are not in meatspace, such as cans on shelves with the side with nutrition information and ingredients list facing forward, rather than  the side with the pretty pictures.  Imagine a shelf of soft drinks arranged by flavor instead of brand-family.  A precursor to this could be virtual store catalogs, which could be more or less simple SQL queries with little or no extra processing.

'virtualworker:key,volunteer,desc_

'depreciation:key,volunteer,desc,life,salvage,method'

'ndb:key,desc,ndb'

'function:key,desc'

'functionpart:key,function,desc_

pubwan output

shopping list generator

virtual store generator

virtual business plan generator