# Perplexed by Quality: A Perplexity-based Method for Adult and Harmful Content Detection in Multilingual Heterogeneous Web Data arXiv: https://arxiv.org/abs/2212.10440 ยท HTML: https://mlsanity.com/paper/2212.10440/ ## Key facts - Cited by 22 papers in the map (data through 14 Oct 2026). - Placed in Safety alignment, refusal and guardrails, one of 12 directions in LLM safety and security. - Builds on 19 of its 27 references that are in the map. - First posted on arXiv 20 Dec 2022. ## Placed in - LLM safety and security / [Safety alignment, refusal and guardrails](https://mlsanity.com/safety/safety-alignment-refusal-and-guardrails/) ## Builds on (most cited first, 19 of 19) - [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017) - [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018) - [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020) - [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019) - [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://mlsanity.com/paper/1910.01108/) (2019) - [The Pile: An 800GB Dataset of Diverse Text for Language Modeling](https://mlsanity.com/paper/2101.00027/) (2020) - [BLOOM: A 176B-Parameter Open-Access Multilingual Language Model](https://mlsanity.com/paper/2211.05100/) (2022) - [Bag of Tricks for Efficient Text Classification](https://mlsanity.com/paper/1607.01759/) (2016) - [CCNet: Extracting High Quality Monolingual Datasets from Web Crawl Data](https://mlsanity.com/paper/1911.00359/) (2019) - [HateXplain: A Benchmark Dataset for Explainable Hate Speech Detection](https://mlsanity.com/paper/2012.10289/) (2020) - [Quality at a Glance: An Audit of Web-Crawled Multilingual Datasets](https://mlsanity.com/paper/2103.12028/) (2021) - [Advances in Pre-Training Distributed Word Representations](https://mlsanity.com/paper/1712.09405/) (2017) - [Towards a Cleaner Document-Oriented Multilingual Crawled Corpus](https://mlsanity.com/paper/2201.06642/) (2022) - [A Monolingual Approach to Contextualized Word Embeddings for Mid-Resource Languages](https://mlsanity.com/paper/2006.06202/) (2020) - [Deep Learning for Hate Speech Detection: A Comparative Study](https://mlsanity.com/paper/2202.09517/) (2022) - [Detecting Harmful Online Conversational Content towards LGBTQIA+ Individuals](https://mlsanity.com/paper/2207.10032/) (2022) - [Hate Speech Classification Using SVM and Naive BAYES](https://mlsanity.com/paper/2204.07057/) (2022) - [Finding Pareto Trade-offs in Fair and Accurate Detection of Toxic Speech](https://mlsanity.com/paper/2204.07661/) (2022) - [Exploring Hate Speech Detection with HateXplain and BERT](https://mlsanity.com/paper/2208.04489/) (2022) ## Built on by (most cited first, 22 of 22) - [LLMDet: A Third Party Large Language Models Generated Text Detection Tool](https://mlsanity.com/paper/2305.15004/) (2023) - [Multi-Task Contrastive Learning for 8192-Token Bilingual Text Embeddings](https://mlsanity.com/paper/2402.17016/) (2024) - [Data Contamination Report from the 2024 CONDA Shared Task](https://mlsanity.com/paper/2407.21530/) (2024) - [An Empirical Study on Cross-lingual Vocabulary Adaptation for Efficient Language Model Inference](https://mlsanity.com/paper/2402.10712/) (2024) - [On Catastrophic Inheritance of Large Foundation Models](https://mlsanity.com/paper/2402.01909/) (2024) - [Towards Safer Pretraining: Analyzing and Filtering Harmful Content in Webscale datasets for Responsible LLMs](https://mlsanity.com/paper/2505.02009/) (2025) - [Guardians and Offenders: A Survey on Harmful Content Generation and Safety Mitigation of LLM](https://mlsanity.com/paper/2508.05775/) (2025) - [FBS: Modeling Native Parallel Reading inside a Transformer](https://mlsanity.com/paper/2601.21708/) (2026) - [CodeShell Technical Report](https://mlsanity.com/paper/2403.15747/) (2024) - [Symmetric Dot-Product Attention for Efficient Training of BERT Language Models](https://mlsanity.com/paper/2406.06366/) (2024) - [Swift Cross-Dataset Pruning: Enhancing Fine-Tuning Efficiency in Natural Language Understanding](https://mlsanity.com/paper/2501.02432/) (2025) - [GeistBERT: Breathing Life into German NLP](https://mlsanity.com/paper/2506.11903/) (2025) - [REIS: A High-Performance and Energy-Efficient Retrieval System with In-Storage Processing](https://mlsanity.com/paper/2506.16444/) (2025) - [SindBERT, the Sailor: Charting the Seas of Turkish NLP](https://mlsanity.com/paper/2510.21364/) (2025) - [Beyond Length: Quantifying Long-Range Information for Long-Context LLM Pretraining Data](https://mlsanity.com/paper/2510.25804/) (2025) - [Oasis: Data Curation and Assessment System for Pretraining of Large Language Models](https://mlsanity.com/paper/2311.12537/) (2023) - [Toxic language detection: a systematic review of Arabic datasets](https://mlsanity.com/paper/2312.07228/) (2023) - [CMHG: A Dataset and Benchmark for Headline Generation of Minority Languages in China](https://mlsanity.com/paper/2509.09990/) (2025) - [Efficient Transformer Knowledge Distillation: A Performance Review](https://mlsanity.com/paper/2311.13657/) (2023) - [PortBERT: Navigating the Depths of Portuguese Language Models](https://mlsanity.com/paper/2606.02100/) (2026) - [Data Processing for the OpenGPT-X Model Family](https://mlsanity.com/paper/2410.08800/) (2024) - [Multilingual Encoder Knows more than You Realize: Shared Weights Pretraining for Extremely Low-Resource Languages](https://mlsanity.com/paper/2502.10852/) (2025) ## Abstract (arXiv) As demand for large corpora increases with the size of current state-of-the-art language models, using web data as the main part of the pre-training corpus for these models has become a ubiquitous practice. This, in turn, has introduced an important challenge for NLP practitioners, as they are now confronted with the task of developing highly optimized models and pipelines for pre-processing large quantities of textual data, which implies, effectively classifying and filtering multilingual, heterogeneous and noisy data, at web scale. One of the main components of this pre-processing step for the pre-training corpora of large language models, is the removal of adult and harmful content. In this paper we explore different methods for detecting adult and harmful of content in multilingual heterogeneous web data. We first show how traditional methods in harmful content detection, that seemingly perform quite well in small and specialized datasets quickly break down when confronted with heterogeneous noisy web data. We then resort to using a perplexity based approach but with a twist: Instead of using a so-called "clean" corpus to train a small language model and then use perplexity so select the documents with low perplexity, i.e., the documents that resemble this so-called "clean" corpus the most. We train solely with adult and harmful textual data, and then select the documents having a perplexity value above a given threshold. This approach will virtually cluster our documents into two distinct groups, which will greatly facilitate the choice of the threshold for the perplexity and will also allow us to obtain higher precision than with the traditional classification methods for detecting adult and harmful content.