Comparing Sequential and Parallel Word Counting Methods in Finding Commonly Used Words by Different Authors

摘要

Parallel computing is exposing to everyday life and becomes a target for parallelism experiments. This paper is a compression between sequential and parallel methods in word counting to find the commonly used words by different authors in a selected public dataset that includes around 500,000 quotes. Python multiprocessing library with a pool of worker processes was used for this purpose. The sequential process (Sequential Method-SM) is grouped into sub-processes "operations" to evaluate the execution time in each operation, and then the operation that consumes the largest portion of execution time is parallelized with multiprocess pool function (Partially Parallelized Method-PPM). After that, the whole process is parallelized to evaluate the overall performance (Fully Parallelized Method-FPM). Results show improved efficiency in the overall parallelized process by 48%, and the most used words by selected authors appear relevant to the authors' profession.