An automated methodology for rapid information extraction from large drilling datasets
Abstract
Extracting information and knowledge from large datasets often takes a significant amount of time in collecting, cleaning and processing the data. This process, from data curation to data interpretation can last from a couple of weeks to several months. Therefore, a structured methodology is developed using concepts such as spider bots and storyboarding to rapidly extract meaningful information from drilling datasets. Three categories of spider bots are identified: cleansing bots, processing bots and indexing bots. These bots efficiently (1) cleanse raw data that may be structured, semi-structured or unstructured, (2) process the cleansed data, and then (3) create index tables so that information can be efficiently retrieved. Next, the storyboarding concept is used to construct a series of visualizations from the information categorized and indexed in the database. Lastly, depending on the question that needs to be answered from the data in the database, a visual report, which contains a summary table and a set of graphs, are generated and presented to the end user. Now, a process that used to take weeks or even months when done manually only takes seconds to generate and present an answer. The method and its effectiveness in rapidly retrieving information from large datasets is demonstrated on a field dataset consisting of five wells on a drilling pad.