Adjustments of the report.

Adjustments of the report.

Adjustments of the report.

More ameliorations of the report.

Report rectified. Detailed results available alongside the report

classify(path) exposed as top level interface.

Minor changes. Software Heritage cited.

Report completed with new results on new database with 374 languages.

Report completed. Old results took off, waiting for new results.

Section 7 almost completed. New language lists. Checker bugs fixed.

Section 6 & 7 almost finished.

Section 6 started. More figures added.

Add classification of file to class CNN.

A manual checking tool for Software Heritage.

Section 5 adjusted and enriched. Figures added.

Section 4 almost completed, Section 5 adjusted.

Metric adjustment and new sections. (Incomplete sections marked with "Ongoing".

Model validity metric migrated to F1 from recall. More sections of the report.

Half of Section 4, start of Section 5

Progression on the report.

First commit of report.

New experiments for a smaller language set.

Minor fixes.

New tokenizer for word-level classification.

A little bit of unsupervised clustering.

N-grams frequency method updated to use .csv dataset.

cnn.py: training on Spark; new graphs of results.

New results of word-level ConvNet method.

cnn_w.py: strings and numbers replaced by special token.

Minor fixes.

cnn.py: import updated.

Word-level ConvNet added. Early stopping added to character-level ConvNet.

Local path issue fixed.

Requirements updated.

import update.

CNN: minor fixes.

CNN: minor fixes.

CNN: new interface of command line.

CNN: minor fixes

Complete CNN trainer and tester.

cnn.py: new training function

Guesslang approach.

Two new approaches: Multinominal Naive Bayes and ConvNet

Make tests interruptible between classes.

Add n-grams approach with probabilities.

Benchmark of average process time.

Visualisation of test results.

Improvement for faster tests.

Initial version of baseline method: letters n-grams with frequency distance.

Add a simple test code. Minor bugs in utils fixed.

Add n-gram model training.

Minor fixes.

Lister fixed, clone into 2-level folders.

Language list from Linguist introduced. Add other repository lists.

Add an integrated script for dataset construction.

Better lists of repositories.

Move scripts to 'scripts' folder.

Fix the issue when Linguist fails to dump JSON file.

Change lister to new lister using Github API.

Script for classifying files by language

Add scripts for listing and cloning repositories.

Move review to docs

import template from swh-py-template (init-py-repo)

Add review of selected articles and tools.