Plagiarism-Detector

A command-line program that performs plagiarism detection using a N-tuple comparison algorithm allowing for synonyms in the text

Command to compile the code.

• Compile with: clang++ -std=c++11 detectorMain.cpp detector.cpp -g -o main.exe –Wall
• Run with: ./main.exe sysn_file_name file1_name file2_name N-size

Assumptions:

• Periods, commas etc. are touching
• There is a space after each synonym (including the last in each line)
• Capitalization has no effect
• “HeLLo”= “hello”
• Tuple is only considered as matching if and only if all the words are same (or synonym) in the exact order
• "for a run" != "run for a”

Approach:

• I created a Detector class, that stores all the words from each file, the synonyms as well as the map of the tuples from file 2.
• Handle all the I/O in the main code
• Outline of Code:
  • Read-in all the files and store comparing files as vector
  • Use a map (Map<string, string>) to store the synonyms:
  • First word is a word that can be replaced with a synonym, second is the default word to replace it
    • Ex. [“run”, “jog”, “sprint”]  <“run” “run”, > <“jog” ,“run”> <“sprint”, “run”>
    • Run is replaced by run, jog is replaced by run, sprint is replaced by run;
    • now the synonyms are taken care of
  • Iterate over both vector < string > and find word in synonyms map ( O-log(n) ) and if found replace with default word
  • Store the tuple of N-Size of input file 2 in an unordered_map <string, int> where string is the tuple and the int is the frequency
  • Frequency is not utilized in this current code, but it can have future applications
  • Iterate over input file 1 and compare N-sized tuples to map of file 2 tuples
    • If match, count variable is incremented
    • Total number of file 1 tuples is recorded in variable total
• Output percentage which is calculated by count/total *100

Limitations:

• Sentence restructuring is not accounted for, since order of tuple is important
• Active vs passive sentence structure will not be detected
• There are only a limited number of word delimiters