Alert

Abusive Language Detection in Online Conversations by Combining Content- and Graph-Based Features

• Copyright 2018-20 Noé Cécillon

Alert is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation. For source availability and license information see LICENCE

• Lab site: http://lia.univ-avignon.fr/
• GitHub repo: https://github.com/CompNet/Alert
• Contact: Noé Cécillon noe.cecillon@univ-avignon.fr

Description

This software was designed to detect abusive messages in online conversations. It is a complete reimplementation and a much extended version of the software used in [PLDL'17, PLDL'17a, PLDL'17b, PLDL'18, PLDL'19]. Two main approaches are implemented: a content-based approach and a graph-based approach, which can also be used jointly. This software was used in C'19, CLDL'19, CLDL'20, C'24.

If you use this software or the related data, please cite reference[CLDL'19] bibtex @InProceedings{Cecillon2019, author = {Cécillon, Noé and Labatut, Vincent and Dufour, Richard and Linarès, Georges}, title = {Abusive Language Detection in Online Conversations by Combining Content- and Graph-based Features}, booktitle = {International Workshop on Modeling and Mining Socia-Media Driven Complex Networks}, year = {2019}, volume = {2}, series = {Frontiers in Big Data}, pages = {8}, doi = {10.3389/fdata.2019.00008}, }

Data

This software was applied to a corpus of chat messages written in French, whih unfortunately cannot be published due to legal matters [PLDL'17, PLDL'17a, PLDL'17b, PLDL'18, PLDL'19, C'19, CLDL'19]. The conversational graphs extracted from these messages are publicly available on Zenodo, though, and can be used by this softwaree. It was also applied to some public data: a corpus of Wikipedia conversations annotated for 3 types of abuse [CLDL'20], which is also available on Zenodo.

Organization

Here are the folders composing the project: * Folder content-based: contains the source code of the content-based approach. * Folder Features: contains the scripts to compute features. * bad_words.txt: the static list of french badwords used in [CLDL'19]. * features.txt: the list of all available features. * Folder graph-based: contains the source code of the graph-based approach. * Folder Features: contains the scripts that compute the features. * Folder train-dev-test: contains the train, development and test splits that we used in [CLDL'20].

Use

1. For the content-based approach, run the main script content-based/main.py with the following arguments:
   • annotations: Path to the file containing annotations such as annotations_attack.csv on Zenodo.
   • messagesdir: Path to the directory containing all conversation files.
   • train: Path to the file containing Ids of all messages in train split. Examples are available in train-dev-test folder.
   • test: Path to the file containing Ids of all messages in test split. Examples are available in train-dev-test folder.
   • classifier: The type of classifier to use. Only SVM is currently available.
   • features: Path to the file containing the subset of features to use. Use features.txt for the full feature set.
2. For graph-based approach, run the main script graph-based/main.py with the following arguments.
   • annotations: Path to the file containing annotations such as annotations_attack.csv on Zenodo.
   • messagesdir: Path to the directory containing all conversation files.
   • train: Path to the file containing Ids of all messages in train split. Examples are available in train-dev-test folder.
   • test: Path to the file containing Ids of all messages in test split. Examples are available in train-dev-test folder.
   • classifier: The type of classifier to use. Only SVM is currently available.
   • window-size: The size of window to use for the weight update.
   • directed: To use directed graphs.

References

• [PLDL'17] É. Papegnies, V. Labatut, R. Dufour, and G. Linarès. Detection of abusive messages in an on-line community, 14ème Conférence en Recherche d'Information et Applications (CORIA), Marseille, FR, p.153–168, 2017. doi: 10.24348/coria.2017.16 - ⟨hal-01505017⟩
• [PLDL'17a] É. Papegnies, V. Labatut, R. Dufour, and G. Linarès. Graph-based Features for Automatic Online Abuse Detection, 5th International Conference on Statistical Language and Speech Processing (SLSP), Le Mans, FR, Lecture Notes in Artificial Intelligence, 10583:70-81, 2017. doi: 10.1007/978-3-319-68456-7_6 - ⟨hal-01571639⟩
• [PLDL'17b] É. Papegnies, V. Labatut, R. Dufour, and G. Linarès. Détection de messages abusifs au moyen de réseaux conversationnels, 8ème Conférence sur les modèles et lánalyse de réseaux : approches mathématiques et informatiques (MARAMI), La Rochelle, FR, 2017. ⟨hal-01614279⟩
• [PLDL'18] É. Papegnies, V. Labatut, R. Dufour, and G. Linarès. Impact Of Content Features For Automatic Online Abuse Detection, 18th International Conference on Computational Linguistics and Intelligent Text Processing (CICling 2017), Budapest, HU, Lecture Notes in Computer Science, 10762:153–168, 2018. doi: 10.1007/978-3-319-77116-8_30 - ⟨hal-01505502⟩
• [PLDL'19] É. Papegnies, V. Labatut, R. Dufour, and G. Linarès. Conversational Networks for Automatic Online Moderation, IEEE Transactions on Computational Social Systems, 6(1):38–55, 2019. doi: 10.1109/TCSS.2018.2887240 - ⟨hal-01999546⟩
• [C'19] N. Cécillon. Exploration de caractéristiques d’embeddings de graphes pour la détection de messages abusifs, MSc Thesis, Avignon Université, Laboratoire Informatique d'Avignon (LIA), Avignon, FR, 2019. ⟨dumas-04073337⟩
• [CLDL'19] N. Cécillon, V. Labatut, R. Dufour & G. Linarès. Abusive Language Detection in Online Conversations by Combining Content- and Graph-Based Features, IAAA ICWSM International Workshop on Modeling and Mining Socia-Media Driven Complex Networks (Soc2Net), Munich, DE, Frontiers in Big Data 2:8, 2019. doi: 10.3389/fdata.2019.00008 - ⟨hal-02130205⟩
• [CLDL'20] N. Cécillon, V. Labatut, R. Dufour & G. Linarès. WAC: A Corpus of Wikipedia Conversations for Online Abuse Detection, 12th Language Resources and Evaluation Conference (LREC), Marseille, FR, p.1375–1383, 2020. Conference version - ⟨hal-02497514⟩
• [C'24] N. Cécillon. Combining Graph and Text to Model Conversations: An Application to Online Abuse Detection, PhD Thesis, Avignon Université, Laboratoire Informatique d'Avignon (LIA), Avignon, FR, 2024. ⟨tel-04441308⟩