Text Error Correction System

Aidan Dyga, Ainslee Plesko, Grant Anderson, Issei Hasegawa

September 1, 2025

Introduction

• Goal: Build a Text Error Correction System
• Input: Original (un-mutated) text file
• Output: Encoded text that can detect errors after mutation/transmission
• Motivation: Ensure data integrity over unreliable channels

Problem

• Problem Statement:
  • Data can be corrupted during transmission/storage
  • Need a way to detect (and possibly correct) errors in text
• Challenges:
  • Random mutations (bit flips, character changes)
  • Efficient encoding/decoding

Example Solution

• Encode each line
  • Every line is duplicated and encoded in a more secure manner to prevent mutation
• Message Decoding
  • After the transfer each line is decoded for comparison
• Line Verification
  • Each line is compared with its encoded variant and corrected where necessary

Python Implementation

Why is this a Tractable Problem?

• This problem is Tractable because it can be solved efficiently in linear time with respect to the input size.
• 1. Linear Time Complexity
  • The system runs in time proportional to the input size (O(n·L)), so it scales efficiently.
• 2. Local Verification
  • Errors are detected and corrected per line using checksums and redundancy, avoiding complex global computation.
• 3. Feasible Error Model
  • With simple mutation errors (like single-copy corruption), correction is straightforward and does not require intractable algorithms.

Limitations

• 1. Homophones
  • The program may not be able to correct words that are spelled correctly but used in the wrong grammatically context.
  • Because they are spelled correctly, they may not get flagged by the checker.
  • Ex: ‘their’ vs ‘there’
• 2. Numerical Errors
  • Because the corrector is based on textual errors, like spelling, numerical differences may not be caught.
  • Ex: ‘1889’ vs ‘1989’
• 3. Incorrect Proper Nouns
  • Unless a proper noun is included within the program’s dictionary, it migt get flagged as an error.
  • A mispelled proper noun that spells as another word may also be missed.
• 4. Structural or Logical Errors
  • Despite its ability to check at a sentence level, it would not be able to check coherance or flow of an entire paragraph.

Conclusion

• Text error correction is essential for reliable communication
• Simple codes can efficiently detect errors
• Tractable due to low computational complexity
• More advanced codes can also correct errors, not just detect