** New pt-arqmath baseline (Apr 18, 2022)

A new baseline for Tasks 1 and 3 implemented in PyTerrier is now available online through GitLab:

	https://gitlab.com/dprl/pt-arqmath

This baseline comes with code to run basic text-based retrieval models (e.g., TF-IDF, BM25) with LaTeX formulas tokenized into text. Evaluation data and code is provided, and people are welcome to use this as a starting point for their own systems. The code is shared under a Mozilla 2.0 license.

*** Tangent-S to ARQMath conversion:
To convert TangentS result to ARQMath result run the following script:
python3 convert_tangets_to_arqmath.py -ldir "latex_representation_v2/" -res TangentS_Res_Task1 -arq arq_math_res.tsv
-ldir: directory in which the latex tsv files are located.
-res: TangentS retrieval results for task 1. (This is the combined results after re-ranking)
-arq: The arqmath format output file.

*** ARQMath to Prim conversion:
To do the evaluation, the arqmath formatted files should be converted to Trec format and them unjudged posts should be removed. 
To do that locate all the arqmath formatted files in a directory, then specify the trec format and prim directory to write the result.
Sample command:
    python get_prim_files_task1.py
    -qre qrel_partial_task1
    -sub "/ARQMath Task 1/All_results/"
    -tre "/ARQMath Task 1/All_Trec/"
    -pri "/ARQMath Task 1/All_Trec_Prim/"

*** Tangent-S and TF-IDF combination:
To have unweighted combination of Tangent-S and TF-IDF results, run the following script on the Trec_formatted result files:
python linear_combination.py -tan TangentS_Res.tsv  -te tf_idf_task1_final.tsv  -res combined_tf_idf_tangents