SAE Update - February 2023 - 25

TECH FOCUS:
AErOSPACE
headings, and how notes are referenced in tables.
Content: Most of the recommendations in this
section can be summed up in one word: consistency.
It is important for standards authors to be consistent
in how they reference properties, elements, data
ranges, and units of measure. Inconsistency results in
more complex AI models, more manual extraction,
more manual verification, and the need to maintain
mapping tables outside of the system.
Drawings and variables: This section covers how to
handle variables shown in drawings and the
importance of using consistent variable names across
standards within a Part Type and Subtype.
The table on the preceding page is a good example
of how machines and humans think differently. Note
the following in the table:
* Multiple values in one column: Several of the
columns have multiple values. some are a different
unit of measure; some represent a range; and the
last column has a mixture of different values,
different units of measure, as well as other symbols
and notes. This can be confusing for a machine to
interpret. It is best to store one value per column so
that each value can be referenced.
* Inconsistent use of punctuation: This table has
columns where numbers in parentheses denote a
different unit of measure. However, parentheses are
also used in some columns to identify a note. How
is a machine supposed to identify when numbers in
parentheses are a different unit of measure versus a
note? Standards authors should be consistent in
how they use punctuation and other written
formats to avoid incorrect interpretations by a
machine.
Given the issues with this table, it would likely require
UPDATE
manual intervention by a human
so that it is digitized correctly-
and this intervention will add cost
and time to your digitization
efforts. However, if the table was
structured differently, it could be
interpreted by a machine without
this added time and cost.
Ultimately, the goal of this
session is to get standards
authors to think differently about
standards. Standards should no
longer be viewed as individual
stand-alone documents.
Standards need to be treated as
data inputs into a database so
that users of digital standards
systems can search across
standards.
Standards authors also need to
understand how consistency will
lead to simpler AI models
required to digitize standards.
And simpler AI models will lead
to faster digitization, with less
manual review and fewer external
mapping tools required to
support the digitization process.
Interested in learning more?
Below are in-person or virtual
options to learn more about how
to write a digital-ready standard:
AeroTech, March 14-16 in Fort
Worth. Or contact Leslie McKay
to arrange a session for your
team. n
February 2023
25
https://www.sae.org/attend/aerotech
SAE Update - February 2023

Table of Contents for the Digital Edition of SAE Update - February 2023
