From hoodel@turtle.mrj.com Fri Sep 23 18:11:30 1994 Status: RO X-VM-v5-Data: ([nil nil nil nil nil nil nil nil nil] ["8453" "Fri" "23" "September" "1994" "17:45:57" "-0600" "Kurt Christensen" "hoodel@mrj.com" nil "210" "Re: Data Element Standardization" "^From:" nil nil "9" nil nil nil nil] nil) Received: from cv3.cv.nrao.edu by fits.cv.nrao.edu (4.1/DDN-DLB/1.5) id AA18194; Fri, 23 Sep 94 18:11:28 EDT Received: from ocfmail.ocf.llnl.gov by cv3.cv.nrao.edu (4.1/DDN-DLB/1.13) id AA11640; Fri, 23 Sep 94 18:11:25 EDT Received: from pierce.llnl.gov by ocfmail.ocf.llnl.gov (4.1/SMI-4.0) id AA01011; Fri, 23 Sep 94 14:44:39 PDT Received: by pierce.llnl.gov (4.1/LLNL-1.18/llnl.gov-05.92) id AA00694; Fri, 23 Sep 94 14:46:08 PDT Return-Path: Received: from turtle.mrj.com by pierce.llnl.gov (4.1/LLNL-1.18/llnl.gov-05.92) id AA00666; Fri, 23 Sep 94 14:45:59 PDT Received: from [192.101.175.102] (mac02.mrj.com [192.101.175.102]) by turtle.mrj.com (8.6.4/8.6.4) with SMTP id RAA26694 for ; Fri, 23 Sep 1994 17:41:32 -0400 Message-Id: <199409232141.RAA26694@turtle.mrj.com> X-Sender: hoodel@mrj.com (Unverified) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" From: hoodel@mrj.com (Kurt Christensen) To: ieee-metadata@llnl.gov Subject: Re: Data Element Standardization Date: Fri, 23 Sep 1994 17:45:57 -0600 Response to paper by Ronald L. Schuldt and Barbara A. Barman I recently came across a paper entitled "Implementing a National Data Element Standardization and Registration Strategy". In reading it I was at once both supportive of and disturbed by it. I got the feeling that your description is application specific, but I can't quite put my finger on why I feel that way. I have made some comments on the prime words and class words you listed. I have left the prime words alone for the most part. I have made several alternate suggestions for class words. I hope you find these comments useful. My contact information is provided below, in case you want me to elaborate on my suggestions. Regards, Kurt Christensen Member of the Technical Staff MRJ, Inc. 10455 White Granite Drive Oakton, VA 22124 (703) 218-0343 (703) 385-4637 (facsimile) hoodel@mrj.com =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= PRIME WORDS: Although I would like to recommend an alternate set of prime words, I cannot. Although somewhat contrary to my philosophical "gut feel", the list provided seems to be most consistent with the legacy of common usage. Asset keep as a non-human "Resource" Same as Human except is non-human. Document keep as "Document" Enterprise keep I would rather it be "Organization" as any succinctly definable group of resources. Dissagree with statement that Enterprise is broader than Organization. I see an Organization comprised of multiple Enterprises. It is just that Enterprise has become the accepted term for this kind of specification. Human keep as an human "Resource" Note: This refers to the Resource itself, not the data or information regarding this resource. Same as Asset, except is human. Law/Rule should be a CLASS WORD - "Guidance". Includes physical or procedural restrictions or recommendations regarding a process. Dissagree with assessment that Law/Rule is broader than Guidance. Process keep as "Process" Product ???? Collides with a proposed CLASS WORD below. Program keep as "Program" Resource any asset or human that is available to a process. State a situation or event where certain criteria are constrained. CLASS WORDS: I'm having trouble with the Date/Time problem. My problem would be solved by making the distinction between absolute time and relative time. The paper seems only to understand a relative time. Thus the need for the date for reference. In terms of one representation, the distinction can be understood as follows. Date = yyyymmdd time = hh:mm:ss.ssss In absolute time this becomes: time - yyyymmddhh:mmss.ssss where hh is specified according to a 24 hour clock. With proper handling, this format would sort properly both numerically, and alphabetically. "Dimension" could refer to any type of value. How about "Length" or "Measure" instead? Typically data will conform to some units standard like mks for meter / kilogram / second, cgs for centimeter / gram / second, or fss for foot / slug / second (allowing inconsistent imperial use of the pound_mass corrected with fps for foot / pound / second). In general we make explicit a standard length, a standard mass and a standard time. Associated temperatures are implicit with Kelvins or Rankines for absolute temperatures and degrees C & degrees F for relative temperatures. All other scientific and engineering measures can be derived from these. Keep in mind that radians are unitless ratios (rate) implying arc_degrees, gradians or whatever. Amount keep wish I had a better word. Could be same as Value. Could be a modifier. What would be the difference between "dollar AMOUNT" and "dolar VALUE"? Angle KILL defined as units of (length / length) * unit_per Use as a modifier for Value. Area KILL defined as units of (length * length) Use as a modifier for Value. Behavior specifies how to predict states of a process. Alternatively, this is a Specification of a process. Category keep Code chosen from an enumerated space (Tuple) Code keep same as Name Content specifies the data within a document. Could refer to text Data or a tuple refering to an ordered set of content. "DOCUMENT TEXT" becomes "DOCUMENT CONTENT". As part of a query, "DOCUMENT text CONTENT" as opposed to "DOCUMENT picture CONTENT" returns different data. Coordinate KILL should be "Tuple" defined as a list of values Use as a modifier for Tuple. (i.e., ASSET location 2D geographic coordinate TUPLE) Here 2D geographic coordinate implies that the TUPLE will be a longitude, latitute pair. Data any data not otherwise specified Date keep declaration of a particular 24 hour time period Description could be same as text Data Semantic metadata about the object in question. Difference Value derived by subtraction or implied as Relative Dimension KILL should be "Measure" defined in units of (length) (Dimension is used to identify a value from a coordinate Tuple.) Force IGNORE defined as units of ((mass * length) / (time * time)) or for imperial units (force). Use as a modifier for Value. Guidance Includes physical or procedural restrictions or recommendations within or without an Enterprise regarding a Process. Identifier keep same as Name except must be unique Law Guidance describing restrictions that cannot or must not be violated. Mass KILL defined as either (mass) or for imperial units ((force * time * time) / length) Use as a modifier for Value. Measure IGNORE defined as (length) Use as a modifier for Value. Name KILL same as Identifier except not necessarily unique Use as a modifier for text Data. (as in "HUMAN name text DATA"). Pressure IGNORE defined as (mass / ( time * time * length)) or for imperial units (force / (length * length)) If we are going to have Temperature, we might make an argument for Pressure, Albedo, Carbon_Dioxide_Density. Use as a modifier for Value. Product value derived from multiplication of quantities Collides with the prime word "Product". Quantity keep Value in numerical space. Modifiers should include "discrete" and "continuous" (int vs float). Rate keep same as Ratio Typically used as a time derivative. Could also be used as "tax RATE" or "labor overhead RATE". Ratio Value derived from division of quantities (Allows for specification of relative humidity & etc.) Rule Guidance describing recommendations that will not or should not be violated. Alternately, in knowledge based systems, rules may specify behavior. Specification Either a Guidance, or a Behavior. Sum Value derived from addition of quantities. Temperature KILL defined as units of (temperature) Should be a modifier for Value. Additional modifiers should include "absolute" and "relative" (e.g., kelvins vs C). Text KILL defined as free text. Use as a modifier for appropriate class words such as Data. Meaning that the data is represented as a character string. (e.g., "HUMAN title text VALUE"). Time keep defined as units of (time) Modifiers sould include "absolute" and "relative" as above. Tuple an ordered set. Could refer to coordinates or content of a document for example. Value any single Data value. Used for physical quantities, such as length, temperature, pressure, weight & etc. Volume KILL defined as units of (length * length * length) Use as a modifier for Value. Weight KILL defined as units of ((mass * length) / (time * time)) or for imperial units (force). Use as a modifier for Value.