Sullivan_CoastPilotDataStructure_noaa_review_2015

Hello, I am Briana Sullivan. I work in the Data Visualization Lab on the “Chart of the
Future” related projects.
I’d like to discuss the Coast Pilot data structure as it is today and some insights on
where I think it should be and some ideas on how to make it happen.
I’ll follow up with a brief synopsis of the work we are involved in with the various IHO
Working Groups.
Joint Hydrographic Center Annual Review *
July 14-16, 2015

let me first explain, visually, how the data structure is based on the publication…
July 14-16, 2015

The Coast Pilot is set up via Books. Books subdivide the working area into
maintainable sections. So, to get local information you need to know the Book as the
root element
July 14-16, 2015

The chapters unique to each book, subdivide the area represented by the book into
manageable parts. NOTE: each chapter contains one or more nautical chart within it’s
limits (and due to this some information repeats where charts/chapters/books share
a border/overlap)
July 14-16, 2015

Within each book are various chapters, some with data common to the entire book,
most chapters are for specific areas delineated by a specific boundary.
Clearly there are some chapters that don’t follow the same type of formatting...these
will be researched and dealt with at a later time.
Notice this disclaimer on the left that this is “…information that is difficult to display
on a nautical chart.” ...I’ll talk about this a bit later.
July 14-16, 2015

Within each chapter are “elements”. Each element is represented as a paragraph
within the chapter and is numbered. But the elements have types associated with
them such as:
Chapter Header (Orange)
Chart Header (Yellow)
Paragraph Header (Aqua)
Paragraph -> CP_Tags (Blue)
Image (Pink)
These items are typically described in geographic order. On the East Coast of the US,
for example, it is from North to South.
July 14-16, 2015

This is a hierarchal data structure ….just like a tree, where the whole CP is the tree,
each book is a branch, each chapter are leaves on the branch, etc.
July 14-16, 2015

This is what the “tree” looks like when represented with XML. You can see the tags
are exactly as you’d expect the layout of the book!
July 14-16, 2015

With the S-100 standard in the works, technology use common place, and a need for
more useable and accessible data it’s “time to think differently”.
It's probably better not to think of Google Maps (or Nautical Charts) as a
thing like a paper map.
Geographic information systems represent a jump from paper maps like the
abacus to the computer.
A quote from London cartographic historian …
(read quote at top of slide).
A former NASA engineer working at Google lays out the steps to the success of
Google Maps…(read quote at bottom of slide)
Isn’t that the holy grail? Getting something out of this that is “higher quality than the
sum of its parts”?
source: http://www.theatlantic.com/technology/archive/2012/09/how-google-
July 14-16, 2015

builds-its-maps-and-what-it-means-for-the-future-of-everything/261913/
9

We can accomplish this if we can move from a publication-centric data format to a
data-centric one. What do I mean by this?
July 14-16, 2015

It brings me back to that disclaimer I mentioned earlier…(read slide).
OCS is starting the process of thinking differently
July 14-16, 2015

…I recently found this on their web page…(read slide)
But remember...just tagging something with XML doesn’t mean it will allow the kind
of control mentioned here.
Thought and foresight needs to come into play to make sure the mark-up is being
used to maximize data use.
source: http://www.nauticalcharts.noaa.gov/nsd/RD-cpdb.html
July 14-16, 2015

So to demonstrate how “Thinking Differently” could work to maximize the
data…here’s an example paragraph.
Notice how the info in the red outlines could be considered “metadata” to this
chaper/book. The Chart headings would be irrelevant in a world of charts based on
features. Now the line is blurred between charts and boundaries that were so
prevalent in the past.
The XML colored in black is directly from paragraph 5 (the green highlight). Basically, I
read the paragraph and attempted to take everything I could and “tag” and organize
it. But, since I’m focused on the feature, I’m stopping at the first one I see, “St. Croix
River”. Then I search the rest of the chapter for “St. Croix River”. My goal is to find
and group all info about the “St. Croix River” that I can. The orange text is just
that…info from the rest of the chapter about the St. Croix River. Wow, how great is it
to be able to find out all about this river in one location!
Doing this exercise brought up quite a few other questions in quality control with
consistency, formatting, and organization. Like:
1) Why isn’t the first instance of “St. Croix River” (5) in bold? Instead it is bold in
July 14-16, 2015

(104).
2) Principal entrance is listed in (5) as “Head Harbour Passage”, yet elsewhere in the
chapter “Friar Roads” is also a principal entrance
3) what is the determining factor to having the geo-referenced links?
4) What aren’t all the occurances within the text (i.e. Whitlocks Mill) all geo-
referenced? I should be able to click on any one of those feature names and get
the same result.
13

the current XML (publication-centric) mark-up...comparing this time against the data-
centric mark-up example on the right.
Note: the St. Croix River (the green box) is inside the <paraText> tag hidden amongst
a narrative string of words. Not easy to quickly find or extract.
Granted...it’s a LOT of work! But the uses for how the data is marked up on the right
is many times more useful than the original.
July 14-16, 2015

For those of you who don’t like to look at code, another way to view and understand
this would be like filling out a form online…
The example on the left allows for a narrative - a rich textual description. Easy for
input purposes, but just not easy to process, discover and use on the output.
The example on the right, sure it’s more complicated to set up, it takes longer to fill
out...BUT, it forces the input to be formatted, tagged, organized, and verified on
submit, which means there is no more/very little work to do on the output...
This is a great example of how to standardize the data content.
July 14-16, 2015

Yet another way to view this data structure (for those of you that like pictures!)
This is the direction of the S-100…Feature-based data…
it can work for the Coast Pilot/Sailing Directions too…in fact it should work in
conjunction with S-101 (the ENC) data.
Everything related to a specific physical feature (S-126 - the physical environment)
should be able to reference the S-101 ENC utilizing the overlap and simply linking the
features together. Remember the goal to reduce redundant work and work towards
“harmonization” with other related specifications.
For the most part the S-126 is basically describing the S-101 feature….the textual
description of the physical feature as it relates to navigation.
July 14-16, 2015

Remember, we need to keep in mind the end result, how will the data be used? It is
(or should) be there to serve us...not us be slave to it.
1. (or better would be “winding” since “windy” in the dictionary is “wind swept”
not “twisty”...but that’s another presentation!)
July 14-16, 2015

Of course, the biggest problem in this whole story is HOW do you get from the
current situation to the dream?
Or even something on the way to the dream? I don’t think the presentation should
end here (even though you may want it to!)…
July 14-16, 2015

….So, how to make it happen? How do you turn a tree into lots of little seeds (for lack
of better analogy)?
And the million dollar question how do you take the seeds and get the tree you are
used to?
July 14-16, 2015

1. Process of reducing redundancy in the DB.
a. Use query string to find distinct Paragraph Header type of elements,
then clean the list manually
i. Get rid of tags, duplicates, make things generic (get rid of state
and regional specific locations in text) - this could be a starter
list for common headers in all of the world's sailing directions --
This is part of the STANDARDIZATION process! (example:
Pilotage for Coast of New Hampshire, just Pilotage since it’s
already in the section about New Hampshire)
1. duplicates in the database reduce headers by 318
entries (from 2258 to 1940),
ii. This is a great exercise to see how consistent the document is
written, how terms/phrases/headers are similar but not the
same.(i.e. using MA vs. Mass, VA vs. Va., NY vs. N.Y., under way
vs. underway, use of caps, use of spaces/punctuation, plural vs.
singular, cross current vs. crosscurrent, small craft vs. small-
craft vs. small-boat)
b. Get rid of added tags and attach as a new field “type” or “paragraph
header type” (Appendix, CFR, sect, CP_INDEX, CP_BOLD, etc.)
July 14-16, 2015

c. Punctuation within tags isn’t necessary.
1. Geo-reference data - how is Tom’s group doing it (regexp automation to start)
According to publication “Fulltext Geocoding Versus Spatial Metadata for
Large Text Archives: Towards a Geographically Enriched Wikipedia”
a. Checking for potential matches: find all possible matches (either use
the xml version of each document or make new version from DB) -
“feature catalogue”
b. match with GNIS table
2. Isolate the the chart numbers using regexp, set up parent/child relationships
via query and create new table of chart numbers to elements.
This set will give everything an associated area even if it doesn’t have a geo-
reference.
3. Compare against the ENC feature catalogue so the tagging can be synced
up/reused
4. Such as Anchorage, Pilotage, etc. Later these can be changed into feature
categories that will comply with S-100 standards.
5. recode without attributes
20

Fewer Attributes: This is an easy interim step that could be done now to help improve
the way the Coast Pilot XML is currently being done.
July 14-16, 2015

The Center has allowed me to become involved with NIPWG and I have presented
them with the ideas in this presentation at their first meeting in June/July 2015. The
Coast Pilot data will primarily become S-126 “The Physical Environment”…the
description of the nautical chart. They are in the process, right now, of determining a
suitable data structure for this kind of data.
We are also helping with the portrayal of the MPA (Marine Protected Areas)
July 14-16, 2015

For the past three years I have been representing the work Colin Ware and Roland
Arsenault have been doing on the portrayal of surface currents. We have provided
them with the research to back up why arrows shaped this way are better for
representing surface currents than other shaped arrows as well as demonstrated how
this can evolve into using streamlines to represent the flow of currents.
July 14-16, 2015

Extra help promised to Julia Powell the Chair of the S-100 working group that
overlaps with work being done for the other two groups.
July 14-16, 2015

Any questions or comments feel free to contact Briana Sullivan:
briana@ccom.unh.edu
July 14-16, 2015

Sullivan_CoastPilotDataStructure_noaa_review_2015

More Related Content

Viewers also liked

Similar to Sullivan_CoastPilotDataStructure_noaa_review_2015

Sullivan_CoastPilotDataStructure_noaa_review_2015