AlasKA is a Python package that reads mnemonics from LAS files and outputs an aliased dictionary of mnemonics and its aliases, as well as a list of mnemonics that cannot be found. It uses three different methods to find aliases to mnemonics: locates exact matches of a mnemonic in an alias dictionary, identifies keywords in mnemonics' description then returns alias from the keyword extractor, and predicts alias using all attributes of the curves.
To install the package first clone the repository to your local machine. Then cd into the repository and create a virtual environment with conda env create -f environment.yml. Next activate the environment with conda activate alaska. From inside the alaska virtual environment and in the alaska top-level directory run pip install . and it will install the latest version of alaska and all the dependencies except for PyTorch. To install PyTorch go to https://pytorch.org/ and select the version of PyTorch that is compatible with your machine. PyTorch is only required if you want to use the pointer generator model to alias mnemonics (~10% of log mnemonics need the pointer generator model).
from alaska import Alias
from welly import Project
path = "testcase.las"
a = Alias()
parsed, not_found = a.parse(path)In this case, parsed is the aliased dictionary that contains mnemonics and its aliases, and not_found is the list of mnemonics that the aliaser did not find. Users can manually alias mnemonics in the not_found list and add them to the dictionary of aliased mnemonics
Parameters of the Alias class can be changed, and the defaults are the following
a = Alias(dictionary=True, keyword_extractor=True, model=False, prob_cutoff=.5)Users can choose which parser to use/not to use by setting the parsers to True/False. The prob_cutoff is the confidence the user wants the predictions made by the pointer generator model parser to have.
Then, the aliased mnemonics can be input into welly as demonstrated below.
from welly import Project
p = Project.from_las(path)
data = p.df(keys=list(parsed.keys()), alias=parsed)
print(data)