« EML and DEIMS - Mapping Mission | Main | Devel snippets »

October 05, 2011

EML and DEIMS - Mapping Attributes

How to treat variable information (called an Attribute in EML) is at the forefront of my mind for a few reasons:

1) Variables can vary widely (ha,ha) among and within data sets which makes the EML specification rather complex.
2) Portions of this complexity are ensconced within the DEIMS metadata structure but much of it is not. I tend to agree with this approach as encasing the entire specification would create a huge proliferation of fields that would rarely be used and would make direct metadata entry into the DEIMS system by non-expert researchers nearly impossible.

On with the show.

EML Specification*

<attributeName> is the official name of an attribute, typically the name of a field in a data table. This is often short and/or cryptic.

<attributeLabel> (optional): is used to provide a less ambiguous or cryptic alternative identification than what is provided in <attributeName>. This content may be used as a column or row header in an HTML display.

<attributeDefinition> gives a precise and complete definition of attribute being documented. It explains the contents of the attribute fully so that a data user can interpret the attribute accurately.

Corresponding DEIMS variable fields

Node Title -> attributeName
Variable Abbreviation (field_attribute_label) -> attributeLabel
Definition (field_var_definition) -> attributeDefinition

This is confusing. Variable Abbreviation maps to attributeLabel although the latter is designed to be the full variable name (i.e., Node Title.)

EML Specification* - yields 5 over-arching variable categories:

<measurementScale> indicates the type of scale from which values are drawn for the attribute. One of the 5 scale types must be used: nominal, ordinal, interval, ratio, or dateTime,

The <nominal> scale is used to represent named categories. Values are assigned to distinguish them from other observations. This would include a list of coded values (e.g. 1=male, 2=female), or plain text descriptions. Columns that contain strings or simple text are nominal. Example: plot1, plot2, plot3.

<ordinal> values are categories that have a logical or ordered relationship to one another, but the magnitude of the differences between the values is not defined or meaningful. Example: Low, Medium, High.

<interval> These measurements are ordinal, but in addition, use equal-sized units on a scale between values. The starting point is arbitrary, so a value of zero is not meaningful. Example: The Celsius temperature scale uses degrees which are equally spaced, but where zero does not represent “absolute zero” (i.e., the temperature at which molecular motion stops), and 20 Celsius is not “twice as hot” as 10 Celsius.

<ratio> measurements have a meaningful zero point, and ratio comparisons between values are legitimate. For example, the Kelvin scale reflects the amount of kinetic energy of a substance (i.e., zero is the point where a substance transmits no thermal energy), and so temperature measured in kelvin units is a ratio measurement. Concentration is also a ratio measurement because a solution at 10 micromolePerLiter has twice as much substance as one at 5 micromolePerLiter.

<dateTime>, is a date-time value from the Gregorian calendar and it is recommended that these be expressed in a format that conforms to the ISO 8601 standard. An example of an allowable ISO date-time is “YYYY-MM-DD”, as in 2004-06-25, or, more fully, as “YYYY-MM-DDThh:mm:ssTZD” (e.g., 1997-07-16T19:20:30.45Z).

Corresponding DEIMS variable fields

The DEIMS implementation is simplified into the following groups:

Quantitative Variable: Interval/Ratio are clumped under ratio

Date Time Variable: dateTime

Text Based Variable: Nominal/Ordinal are clumped under nominal.

- Note: pattern here is /attribute/measurementScale/nominal/nonNumericDomain/enumeratedDomain - are the last two contradictory? Partial answer: possibly not because there is a numericDoman field

EML Specification* - "The and scales require additional tags describing , the , and."


<unit> Units should be described in correct physical units. Terms which describe data but are not units should be used in <attributeDefinition>. For example, for data describing “milligrams of Carbon per square meter”, “Carbon” belongs in the <attributeDefinition>, while the <unit> is “milligramPerMeterSquared”.

Corresponding DEIMS variable fields

Unit (field_attribute_unit) -> Unit within a customUnit tag

Notes to follow-up on:

Code-definition doesn't show up in the EML output. views-bonus-eml-export-eml.tpl.php indicates it should appear as
attribute/measurementScale/nominal/nonNumericDomain/enumeratedDomain/code
attribute/measurementScale/nominal/nonNumericDomain/enumeratedDomain/defintion

Although it seems this would be correct

attribute/measurementScale/nominal/nonNumericDomain/enumeratedDomain/codeDefition/code
attribute/measurementScale/nominal/nonNumericDomain/enumeratedDomain/codeDefintion/defintion

*EML Best Practices Working Group. EML Best Practices for LTER Sites V2.0. August 1st, 2011. http://im.lternet.edu/sites/im.lternet.edu/files/emlbestpractices-2.0-FINAL-20110801_0.pdf

Posted by kkwaiser at October 5, 2011 11:19 AM

Trackback Pings

TrackBack URL for this entry:
http://mblog.lib.umich.edu/mt-bin/mt-tb.cgi/2256

Listed below are links to weblogs that reference EML and DEIMS - Mapping Attributes:

» Click para ir a la página web from Click para ir a la página web
Data Discussions: EML and DEIMS - Mapping Attributes [Read More]

Tracked on February 28, 2014 08:45 PM

» web site from web site
Data Discussions: EML and DEIMS - Mapping Attributes [Read More]

Tracked on March 19, 2014 12:02 PM

» tory burch reva flats from tory burch reva flats
Data Discussions: EML and DEIMS - Mapping Attributes [Read More]

Tracked on March 27, 2014 08:30 PM

» christian louboutin glitter shoes from christian louboutin glitter shoes
Bruk åttende periode affordable christian louboutin shoes for mer nettbasert utdanning [Read More]

Tracked on March 27, 2014 09:49 PM

» shit bricks from shit bricks
Data Discussions: EML and DEIMS - Mapping Attributes [Read More]

Tracked on March 27, 2014 10:51 PM

» www.candycaneapps.com from www.candycaneapps.com
Data Discussions: EML and DEIMS - Mapping Attributes [Read More]

Tracked on April 6, 2014 03:39 AM

Comments

Login to leave a comment. Create a new account.