• Nikhat Dharani (ndharani@college)
• Pin-Wen Wang (pin-wen_wang@college)

The Rabin-Karp Algorithm is an implementation of exact string matching that uses a rolling hash to find any one set of pattern strings in a text.

The Rabin-Karp algorithm is used in detecting plagiarism because given a pattern and a source of texts, the algorithm can quickly search through papers for patterns from the source material.

In this project, we implement serial and parallel versions of the Rabin-Karp algorithm, loosely based on the paper: "Parallelized Rabin-Karp Method for Exact String Matching" by Brodanac, P. Budin, L, Jakobovic, D. http://ieeexplore.ieee.org/xpl/abstractAuthors.jsp?arnumber=5974088, and we explore different methods of parallelizing the Rabin-Karp algorithm and analyze our findings here: http://people.seas.harvard.edu/~pwwang/index.html

###Approaches: ### There are a number of approaches to parallelizing the Rabin-Karp algorithm that we will design and analyze. We can divide the corpus of texts, and/or the pattern among processes and run the algorithm on the corpus in real-time (with the rolling hash) or preprocessed. Inspired by the original Rabin-Karp algorithm, we implement the serial and parallel versions as well as attempt our own parallel version using prehashed corpuses.

• Takes the Rabin-Karp algorithm and parallelizes the rolling hash using MPI.

• Prehash the corpus of texts that will be searched through by hashing each word using MapReduce, and search on the pre-processed corpus of texts for patterns.

• MapReduce (MRJob)
• MPI (mpi4py)
• Python (sys, string, numpy, itertools, pandas, subprocess, time, operator)

The above packages should already be installed on the resonance.seas cluster for CS205. On the resonance node, module load courses/cs205/2013 will load the modules automatically.

• RabinKarpSerial.py: Serial implementation of the Rabin-Karp algorithm on a corpus of texts
• RabinKarpParallel.py: MPI-master-slave parallel implementation of the Rabin-Karp algorithm (as described in the paper) on a corpus of texts
• mpirka.py: single text MPI-send-recv parallel implementation of the Rabin-Karp algorithm
• RabinKarpEmbarParallel.py: MPI-master-slave implementation of running the Rabin-Karp algorithm on a corpus of texts - embarassingly parallel method of sending each processor an entire source text file to run the Rabin-Karp algorithm
• MRhash_word.py: Prehashes a source text by word using MapReduce and divides the hashed text into X chunks
• MRhash.py: Obsolete mapper/reducer. This file is used because a function defined in it (letsHash) is still relevant to other code.
• mpiRK_chunkCorpus.py: [prehashed corpus] single text parallel implementation of Rabin-Karp algorithm where each process searches for pattern in one chunk of prehashed corpus text.
• mpiRK_chunkMultCorpus.py: [prehashed corpus] multiple texts parallel implementation of Rabin-Karp algorithm where each process searches for pattern in one chunk of prehashed corpus texts.
• MasterSlave_chunkCorpus.py: [prehashed corpus] single text is chunked into small pieces and dynamically distributed to slaves.
• MasterSlave_chunkMultCorpus.py: [prehashed corpus] multiple texts are individually chunked into small pieces and dynamically distributed to slaves.
• regroupText2.py: Reshapes text into K pieces and hashes text using MRhash_word.py, assuming m identical words=plagiarism. Note: makes a command line call.

Inside the directory Hash-by-Char you can run the serial and parallel code by running the following:

• RabinKarpSerial.py: serial version that hashes in real time

Usage: python RabinKarpSerial.py [file w/ corpus text names] [pattern text]
Example: python RabinKarpSerial.py filenames.txt multipattern.txt

• RabinKarpParallel.py: parallel version that hashes in real time

Usage: mpiexec -n [# of processes] python RabinKarpParallel.py [file w/ corpus text names] [pattern text]
Example: mpiexec -n 20 python RabinKarpParallel.py filenames.txt multipattern.txt

Inside the directory Hash-by-Word you can run the serial and parallel code by running the following:

• prehashedSerialRK.py: serial version using prehashed corpus

Usage: python prehashedSerialRK.py [text file hashed w/ regroupText2.py] [pattern file (unhashed)]
Example: python prehashedSerialRK.py txt/english_smallerHashed_2000chunks.txt multipattern50.txt

• MasterSlave_chunkCorpus.py: parallel version w/ master/slave framework using prehashed corpus

Usage: mpiexec -n [# of processes] python MasterSlave_chunkCorpus.py [text file hashed w/ regroupText2.py] [pattern file (unhashed)]
Example: mpiexec -n 20 python MasterSlave_chunkCorpus.py txt/english_smallerHashed_2000chunks.txt multipattern50.txt

• MasterSlave_chunkMultCorpus.py: master/slave framework using prehashed corpus w/ multiple texts (filenames)

Usage: mpiexec -n [# of processes] python MasterSlave_chunkMultCorpus.py [file w/ list of hashed files] [pattern file (unhashed)]
Example: mpiexec -n 20 python MasterSlave_chunkMultCorpus.py txt/hashedfilenames.txt multipattern50.txt

• mpiRK_chunkCorpus.py: mpi parallel version w/ #chunks of corpus = #processes, using prehashed corpus

Usage: mpiexec -n [# of processes] python mpiRK_chunkCorpus.py [text file hashed w/ regroupText2.py] [pattern file (unhashed)]
Example: mpiexec -n 20 python mpiRK_chunkCorpus.py txt/english_smallerHashed_20chunks.txt multipattern50.txt

• mpiRK_chunkMultCorpus.py: mpi w/ #chunks of corpus w/ multiple texts (filenames)

Usage: mpiexec -n [# of processes] python mpiRK_chunkMultCorpus.py [file w/ list of hashed files] [pattern file (unhashed)]
Example: mpiexec -n 20 python mpiRK_chunkMultCorpus.py txt/hashedfilenames.txt multipattern50.txt

• regroupText2.py: reshape and hash corpus text. text must be in txt folder.

Usage: python regroupText2.py [filename w/out '.txt'] [# of chunks] [# of words constituting plagiarism]
Example: python regroupText2.py PrideAndPrejudice 20 20

• Louis Mullie: Project Mentor
• Cris Cecka: Course instructor