tsim 0.1.0

A lightweight library for quantifying the similarity between two strings

TSim

A lightweight string similarity quantifier, with support for names

What it does

TSim (Text Similarity) is a quantifier for similarity of strings. It reports the extent of similarity of two strings. TSim accounts for errors like OCR errors, different ordering, abbreviation and more.

Features

• Compensation for OCR errors related to similar looking characters (eg. 5 and S; 0 and O)
• Compensation for over-detection and under-detection of text by OCR systems
• Compensation for different ordering of names (eg. FirstName-LastName and LastName-FirstName)
• Compensation for abbreviation in names (eg. Sanyam Asthana and S. Asthana)

Installation

TSim can be installed via pip:

pip install tsim

You can verify the installation by running:

pip show tsim

Usage

To use TSim, you first need to install it using pip.

To use TSim in a project, you need to import the library in your project using import tsim

Functions

get_confidence()

Returns the confidence/error of similarity of two strings

Parameters:
received_str – The string to compare
expected_str – The string to be expected in the comparison
mode – "c" to return confidence, "e" to return the error

Returns: Confidence (0-1)/Error depending on the mode used

get_abbreviated_confidence()

Returns the confidence of similarity of two strings taking into account abbreviated part of the string

Parameters:
received_str – The string to compare
expected_str – The string to be expected in the comparison
mode – "c" to return confidence, "e" to return the error

Returns: Confidence (0-1)/Error depending on the mode used

get_name_confidence()

Returns the confidence of similarity between two names

Parameters:
received_name – The name to compare
expected_name – The name to be expected in the comparison

Returns: The confidence level of the similarity of the names (0-1)

Technical Details

• Python 3.7+
• Uses an error based confidence system, with different penalties per kind of error.
• Uses a dictionary for OCR character similarity matching

License

MIT License

---