TSSort 0.0.2

A library and tool for implementing TSSort in Python, which allows for the sorting of a list using pairwise comparisons while accounting for noisy data.

This repository contains an implementation of the TSSort algorithm proposed in "TSSort – Probabilistic Noise Resistant Sorting" (Hees et al., 2011).

Expected Input

This module expects a text file with a list of sentences separated by newline and a text file of a list of pairwise comparisons in the form int,int and separated by newline. See the sample data files comparisons.txt and sentences.txt in the linked github for an example. The list of pairwise comparisons refers to the sentence index (starting at 0), wherein the first number is the easier sentence and the second number is the harder sentence.

To run the code:

python3 -m tssort comparisons.txt sentences.txt [-l or -n]

where the comparisons.txt file is your list of comparisons and the sentences.txt file is you list of sentences. Note, the comparisons.txt file must reference the index ordering of sentences in the sentences.txt file.

The -l argument will output the ranked list of the sentences, from easiest to hardest.

The -n argument outputs suggestions for the sentences that should be compared next to ensure the most efficient method of sorting the list.