Skip to main content

The NTCIR-10 Math Converter package converts NTCIR-10 Math dataset and relevance judgements to the NTCIR-11 Math-2, and NTCIR-12 MathIR format.

Project description

Introduction

The retrieval unit in the NTCIR-10 Math task dataset is an arXiv document and the judgement unit in the relevance judgements is an XML element. On the other hand, the retrieval and judgement units in the NTCIR-11 Math-2, and NTCIR-12 MathIR task dataset, and relevance judgements is an arXiv document paragraph. This makes it difficult to use both datasets together in a single evaluation.

NTCIR Math converter is a Python 3 command-line utility that converts the NTCIR-10 Math dataset and relevance judgements to the NTCIR-11 Math-2, and NTCIR-12 MathIR format by splitting the dataset into paragraphs and redirecting the relevance judgements from elements to their ancestral paragraphs. As a result, the NTCIR-10 Math dataset, and relevance judgements can be easily used together with the NTCIR-11 Math-2, and NTCIR-12 MathIR dataset, and relevance judgements in a single evaluation.

Usage

Installing:

$ pip install ntcir10-math-converter

Displaying the usage:

$ ntcir10-math-converter --help
usage: ntcir10-math-converter [-h] --dataset DATASET [DATASET ...]
                              [--judgements JUDGEMENTS [JUDGEMENTS ...]]
                              [--num-workers NUM_WORKERS]

Convert NTCIR-10 Math dataset and relevance judgements to the NTCIR-11 Math-2,
and NTCIR-12 MathIR format.

optional arguments:
  -h, --help            show this help message and exit
  --dataset DATASET [DATASET ...]
                        A path to a directory containing the NTCIR-10 Math
                        dataset, and a path to a non-existent directory that
                        will contain resulting dataset in the NTCIR-11 Math-2,
                        and NTCIR-12 MathIR format. If only the path to the
                        NTCIR-10 Math dataset is specified, the dataset will
                        be read to find out the mapping between element
                        identifiers, and paragraph identifiers. This is
                        required for converting the relevance judgements.
  --judgements JUDGEMENTS [JUDGEMENTS ...]
                        Paths to the files containing NTCIR-10 Math relevance
                        judgements (odd arguments), followed by paths to the
                        files that will contain resulting relevance judgements
                        in the NTCIR-11 Math-2, and NTCIR-12 MathIR format
                        (even arguments).
  --num-workers NUM_WORKERS
                        The number of processes that will be used for
                        processing the NTCIR-10 Math dataset. Defaults to 1.

Converting both a dataset and relevance judgements using 64 worker processes:

$ ntcir10-math-converter --num-workers 64 \
>     --dataset ntcir-10 ntcir-10-converted \
>     --judgements \
>         NTCIR_10_Math-qrels_ft.dat NTCIR_10_Math-qrels_ft-converted.dat \
>         NTCIR_10_Math-qrels_fs.dat NTCIR_10_Math-qrels_fs-converted.dat
Retrieving judged document names, and element identifiers from NTCIR_10_Math-qrels_ft.dat
100%|███████████████████████████████████████████████████████| 1425/1425 [00:00<00:00, 9634.03it/s]
Retrieving judged document names, and element identifiers from NTCIR_10_Math-qrels_fs.dat
100%|███████████████████████████████████████████████████████| 2129/2129 [00:00<00:00, 9671.33it/s]
Processing dataset ntcir-10
Converting dataset ntcir-10 -> ntcir-10-converted/xhtml5
Building a mapping between element identifiers, and paragraph identifiers
100%|████████████████████████████████████████████████████| 100000/100000 [06:45<00:00, 246.50it/s]
Converting relevance judgements NTCIR_10_Math-qrels_ft.dat -> NTCIR_10_Math-qrels_ft-converted.dat
100%|█████████████████████████████████████████████████████| 1425/1425 [00:00<00:00, 252199.81it/s]
Converting relevance judgements NTCIR_10_Math-qrels_fs.dat -> NTCIR_10_Math-qrels_fs-converted.dat
100%|█████████████████████████████████████████████████████| 2129/2129 [00:00<00:00, 291048.96it/s]

Converting only a dataset using 64 worker processes:

$ ntcir10-math-converter --num-workers 64 \
$     --dataset ntcir-10 ntcir-10-converted
Processing dataset ntcir-10
Converting dataset ntcir-10 -> ntcir-10-converted/xhtml5
100%|████████████████████████████████████████████████████| 100000/100000 [07:34<00:00, 220.10it/s]

Converting only relevance judgements using 64 worker processes:

$ ntcir10-math-converter --num-workers 64 \
>     --dataset ntcir-10 \
>     --judgements \
>         NTCIR_10_Math-qrels_ft.dat NTCIR_10_Math-qrels_ft-converted.dat \
>         NTCIR_10_Math-qrels_fs.dat NTCIR_10_Math-qrels_fs-converted.dat
Retrieving judged document names, and element identifiers from NTCIR_10_Math-qrels_ft.dat
100%|███████████████████████████████████████████████████████| 1425/1425 [00:00<00:00, 9539.55it/s]
Retrieving judged document names, and element identifiers from NTCIR_10_Math-qrels_fs.dat
100%|███████████████████████████████████████████████████████| 2129/2129 [00:00<00:00, 9332.81it/s]
Processing dataset ntcir-10
Building a mapping between element identifiers, and paragraph identifiers
100%|████████████████████████████████████████████████████████| 2405/2405 [00:16<00:00, 144.41it/s]
Converting relevance judgements NTCIR_10_Math-qrels_ft.dat -> NTCIR_10_Math-qrels_ft-converted.dat
100%|█████████████████████████████████████████████████████| 1425/1425 [00:00<00:00, 260760.14it/s]
Converting relevance judgements NTCIR_10_Math-qrels_fs.dat -> NTCIR_10_Math-qrels_fs-converted.dat
100%|█████████████████████████████████████████████████████| 2129/2129 [00:00<00:00, 299442.45it/s]

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for ntcir10-math-converter, version 0.1.3
Filename, size File type Python version Upload date Hashes
Filename, size ntcir10_math_converter-0.1.3-py2.py3-none-any.whl (8.8 kB) File type Wheel Python version py2.py3 Upload date Hashes View hashes
Filename, size ntcir10_math_converter-0.1.3.tar.gz (6.9 kB) File type Source Python version None Upload date Hashes View hashes

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN SignalFx SignalFx Supporter DigiCert DigiCert EV certificate StatusPage StatusPage Status page