| |
| author |
Karan Erry
| | title |
Resolving Abbreviations and Domain Terms in Source Code using Documentation
| | abstract |
Hitherto, term resolution techniques have focused on resolving expansions for abbreviations, i.e.
terms that correspond textually with their meanings. However, in many contexts that deal with mathematical
or software variables, the variables do not borrow characters from the meanings they represent.
This means that these variables cannot be expanded using abbreviation-expansion techniques.
Additionally, current term-resolution techniques search for expansions primarily in
source code, largely ignoring a wealth of knowledge contained in the accompanying documentation.
In this paper, I present novel techniques to allow the variables described above to be resolved as well,
using cues inspired by how a human understands the meaning of a piece of code/documentation.
My techniques are designed to search documentation, leveraging natural-language clues.
I also present re-finements to previous papers' acronym-expanding techniques. For all
techniques I achieve recall averaging in the upper third and fourth quartiles.
| | school |
The College of Liberal Arts, Drew University
| | degree |
B.A. (2019)
|
| advisor |
Emily Hill
|
| committee |
Barry Burd
Seung-Kee Lee
|
| full text | KErry.pdf |
| |
