Loader rectification tools:Picklists: closed (and open?)
Develop rules for rectification
#1 Updated by Michael Lee over 19 years ago
I think that this bug deals with the mapping of one set of values into another
set of values. I'm not sure if we want the user to be able to do this from data
on upload (only applicable to NPS and XML -- incl. TurboVeg -- VegBranch takes
cara of this on its own), or if we are to specify values ahead of time (i.e. and
XML doc mapping values onto closed list values). The former sounds a bit
challenging (they could just use VegBranch), and the latter sounds fairly
reasonable (we already are using a similar XML doc to validate values). The
former feature is probably for version 2.0.0; the latter (which this bug
represents, unless otherwise changed later) could be for version 1.0.0.
#2 Updated by Robert Peet about 19 years ago
Not sure what this bug refers to.
Step one is creation of the picklists. I think these are now stable and we
can mark this part as complete.
Step two is to make certain that plots being submitted have enteries match the
appropriate picklists. NOte sure if this is in place yet, but it should be
Gabe's job, so I will reassign this bug to Gabe.
#3 Updated by Gabriel Farrell almost 19 years ago
This is the creation of web ui that allows users to correct fields in the data
they submit when vegbank has a problem with a particular data field(s).
So a user is loading a plot and they have a rocktype in their data that vegbank
does't recognize. A screen pops up asking the user to add one of the acceptable
The picklists are defined in the database schema most easily accessible from
java in the datamodel classes e.g. for a list of acceptable rocktypes
I am unsure if this also refers to doing the same for unrecognized plant or
#4 Updated by P. Anderson almost 19 years ago
- the same rock types, plants, and communities can be named differently
- typos are inevitable
- an interactive rectifier tool is possible
- the tool can make a best guess using a phonetic code (pronunciation)
- guess with a SQL LIKE '%abbreviation%' query too
We could generate phonetic keys from the names of our VegBank data (rocks,
plants, etc.), then simply search that index for any mismatched input to make a
SOUNDEX is a SIMPLE technique devised by Donald Knuth FOR FILING NAMES by the
way they are PRONOUNCED rather than the way they are spelled.
METAPHONE, an algorithm described by Lawrence Philips in the December 1990 issue
of Computer Language, works BETTER than Soundex because it knows the basic rules
of English pronunciation but it uses more storage and computation.
Jakarta has open source Java code for both algorithms.
This tools is totally possible but definitely requires some work to be done right.