{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,17]],"date-time":"2026-10-17T16:21:06Z","timestamp":1792254066204,"version":"4.4.0"},"reference-count":58,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2025,6,19]],"date-time":"2025-06-19T00:00:00Z","timestamp":1750291200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Pediatr."],"abstract":"In online spaces, children are vulnerable to exploitation and sexual predators. Groomers contact minors in online chat rooms with the intent of sexual abuse. This study investigates how new deep learning models compare to traditional machine learning models in detecting grooming conversations and predatory authors. Furthermore, we detect the underlying tones used by predators and explore how these affect detection capabilities. Our goal is to better understand predator tactics and to advance automatic grooming detection in order to protect children in online spaces. The PAN12 chat logs, which contain grooming chat conversations, were used as the dataset for the research. These chat conversations were sorted into sentiments through the DistilBERT classifier based on the predator tone. SVMs and the LLaMA 3.2 1B large language model by Meta were then trained and fine-tuned on the different sentiments. The results measured through precision, recall and F<\/mml:mi><\/mml:mrow>1<\/mml:mn><\/mml:msub><\/mml:math><\/jats:inline-formula> score show that the large language model performs better in grooming detection than traditional machine learning. Moreover, performance differences between the positive and negative sentiment are captured and indicate that positive tone improves detection while negative toned grooming conversations have nuanced patterns that are harder to distinguish from non-grooming. This shows that groomers employ varying strategies to gain access to their victims. Lastly, with an F<\/mml:mi><\/mml:mrow>1<\/mml:mn><\/mml:msub><\/mml:math><\/jats:inline-formula> score of 0.99 and an F<\/mml:mi><\/mml:mrow>0.5<\/mml:mn><\/mml:mrow><\/mml:msub><\/mml:math><\/jats:inline-formula> score of 0.99, the LLaMA 3.2 1B model outperforms both traditional machine learning, as well as previous versions of the large language model in grooming author detection.<\/jats:p>","DOI":"10.3389\/fped.2025.1591828","type":"journal-article","created":{"date-parts":[[2025,6,19]],"date-time":"2025-06-19T05:29:52Z","timestamp":1750310992000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":4,"title":["Comparing machine learning models with a focus on tone in grooming chat logs"],"prefix":"10.3389","volume":"13","author":[{"given":"Leonie","family":"Hamm","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Steve","family":"McKeever","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2025,6,19]]},"reference":[{"key":"B1","article-title":"Data from: Teens, social media and technology 2023","author":"Anderson","year":""},{"key":"B2","doi-asserted-by":"crossref","DOI":"10.18653\/v1\/2021.acl-long.386","article-title":"Early detection of sexual predators in chats","author":"Vogt","year":""},{"key":"B3","first-page":"628","article-title":"Cybergrooming: risk factors, coping strategies and associations with cyberbullying","volume":"24","author":"Wachs","year":"2012","journal-title":"Psicothema"},{"key":"B4","doi-asserted-by":"publisher","first-page":"114808","DOI":"10.1016\/j.eswa.2021.114808","article-title":"Large-scale analysis of grooming in modern social networks","volume":"176","author":"Lykousas","year":"2021","journal-title":"Expert Syst Appl"},{"key":"B5","doi-asserted-by":"publisher","first-page":"301022","DOI":"10.1016\/j.fsidi.2020.301022","article-title":"Detecting child sexual abuse material: a comprehensive survey","volume":"34","author":"Lee","year":"2020","journal-title":"Forensic Sci Int Digit Invest"},{"key":"B6","first-page":"411","article-title":"Child protection on the internet-challenges for criminal law","volume":"14","author":"Gillespie","year":"2002","journal-title":"Child Fam LQ"},{"key":"B7","first-page":"79","article-title":"Online grooming as a manipulative social interaction: insights from textual analysis","volume":"4","author":"Horskykh","year":"2018","journal-title":"Pol J Engl Stud"},{"key":"B8","doi-asserted-by":"publisher","first-page":"287","DOI":"10.1080\/13552600601069414","article-title":"Sexual grooming of children: review of literature and theoretical considerations","volume":"12","author":"Samantha Craven","year":"2006","journal-title":"J Sex Aggress"},{"key":"B9","doi-asserted-by":"publisher","first-page":"946","DOI":"10.21917\/ijsc.2015.0133","article-title":"Supervised machine learning approaches: a survey","volume":"5","author":"Muhammad","year":"2015","journal-title":"ICTACT J Soft Comput"},{"key":"B10","doi-asserted-by":"publisher","first-page":"1","DOI":"10.2352\/ISSN.2470-1173.2016.17.DRR-063","article-title":"Recognizing predatory chat documents using semi-supervised anomaly detection","volume":"28","author":"Ebrahimi","year":"2016","journal-title":"Electron Imaging"},{"key":"B11","doi-asserted-by":"publisher","first-page":"110039","DOI":"10.1016\/j.knosys.2022.110039","article-title":"Online grooming detection: a comprehensive survey of child exploitation in chat logs","volume":"259","author":"Borj","year":"2023","journal-title":"Knowl Based Syst"},{"key":"B12","doi-asserted-by":"crossref","DOI":"10.1109\/SMAP49528.2020.9248461","article-title":"On preprocessing the data for improving sexual predator detection: anonymous for review","author":"Borj","year":""},{"key":"B13","doi-asserted-by":"crossref","DOI":"10.1109\/ETNCC59188.2023.10284950","article-title":"Identifying sexual predators in chats using svm and feature ensemble","author":"Fauzi","year":""},{"key":"B14","article-title":"Data from: Fine-tuning llama 2 large language models for detecting online sexual predatory chats and abusive texts (2023)","author":"Nguyen","year":""},{"key":"B15","doi-asserted-by":"crossref","DOI":"10.1145\/3600211.3604696","article-title":"Protecting children from online exploitation: can a trained model detect harmful communication strategies?","author":"Cook","year":""},{"key":"B16","doi-asserted-by":"publisher","first-page":"220","DOI":"10.1109\/TCIAIG.2015.2424932","article-title":"Detecting predatory behavior in game chats","volume":"7","author":"Cheong","year":"2015","journal-title":"IEEE Trans Comput Intell AI Games"},{"key":"B17","article-title":"Identifying online child sexual texts in dark web through machine learning and deep learning algorithms","author":"Ngo","year":""},{"key":"B18","article-title":"Overview of the international sexual predator identification competition at pan-2012","author":"Inches","year":""},{"key":"B19","doi-asserted-by":"publisher","first-page":"27","DOI":"10.56471\/slujst.v4i.266","article-title":"Sentiment classification: review of text vectorization methods: bag of words, tf-idf, word2vec and doc2vec","volume":"4","author":"Abubakar","year":"2022","journal-title":"SLU J Sci Technol"},{"key":"B20","article-title":"Detecting online grooming by simple contrastive chat embeddings","author":"Borj","year":""},{"key":"B21","article-title":"Simcse: simple contrastive learning of sentence embeddings (2021)","author":"Gao","year":""},{"key":"B22","doi-asserted-by":"publisher","first-page":"108017","DOI":"10.1016\/j.knosys.2021.108017","article-title":"How to take advantage of behavioral features for the early detection of grooming in online conversations","volume":"240","author":"Milon-Flores","year":"2021","journal-title":"Knowl Based Syst"},{"key":"B23","doi-asserted-by":"crossref","DOI":"10.1109\/BIOSIG52210.2021.9548303","article-title":"Detecting sexual predatory chats by perturbed data and balanced ensembles","author":"Borj","year":""},{"key":"B24","doi-asserted-by":"crossref","DOI":"10.1109\/IWBF49977.2020.9107945","article-title":"Ensemble method for sexual predators identification in online chats","author":"Fauzi","year":""},{"key":"B25","doi-asserted-by":"publisher","first-page":"33","DOI":"10.1016\/j.diin.2016.07.001","article-title":"Detecting predatory conversations in social media by deep convolutional neural networks","volume":"18","author":"Ebrahimi","year":"2016","journal-title":"Digit Invest"},{"key":"B26","doi-asserted-by":"crossref","DOI":"10.1145\/3701716.3715528","article-title":"Pushing the limit of llm capacity for text classification","author":"Zhang","year":""},{"key":"B27","article-title":"Attention is all you need","author":"Vaswani","year":""},{"key":"B28","article-title":"Bitfit: simple parameter-efficient fine-tuning for transformer-based masked language-models (2021)","author":"Zaken","year":""},{"key":"B29","article-title":"When scaling meets llm finetuning: the effect of data, model and finetuning method (2024)","author":"Zhang","year":""},{"key":"B30","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1145\/3495162","article-title":"A survey on text classification: from traditional to deep learning","volume":"13","author":"Li","year":"2022","journal-title":"ACM Trans Intell Syst Technol"},{"key":"B31","doi-asserted-by":"publisher","first-page":"6609","DOI":"10.1016\/j.eswa.2015.04.042","article-title":"Example-dependent cost-sensitive decision trees","volume":"42","author":"Bahnsen","year":"2015","journal-title":"Expert Syst Appl"},{"key":"B32","doi-asserted-by":"publisher","first-page":"554","DOI":"10.1016\/j.asoc.2013.08.014","article-title":"Cost-sensitive decision tree ensembles for effective imbalanced classification","volume":"14","author":"Krawczyk","year":"2014","journal-title":"Appl Soft Comput"},{"key":"B33","doi-asserted-by":"crossref","DOI":"10.1109\/RIDE.1997.583715","article-title":"Generalization and decision tree induction: efficient classification in data mining","author":"Kamber","year":""},{"key":"B34","doi-asserted-by":"crossref","DOI":"10.1109\/MASCON51689.2021.9563557","article-title":"Overfit analysis on decision tree classifier for fault classification in damadics","author":"Mahalingam","year":""},{"key":"B35","doi-asserted-by":"crossref","DOI":"10.1109\/ISS1.2019.8908018","article-title":"A study on support vector machine based linear and non-linear pattern classification","author":"Ghosh","year":""},{"key":"B36","article-title":"Large language models: a survey (2024)","author":"Minaee","year":""},{"key":"B37","article-title":"Data from: Google AI gemma open models\u2014Google for developers\u2014Google AI for developers (2024)","year":""},{"key":"B38","article-title":"Data from: Gemma: introducing new state-of-the-art open models (2024)","author":"Banks","year":""},{"key":"B39","article-title":"A survey of large language models (2023)","author":"Zhao","year":""},{"key":"B40","article-title":"Gemma 2: Improving open language models at a practical size (2024)","author":"Gemma Team","year":""},{"key":"B41","article-title":"Data from: Llama 3.2: revolutionizing edge AI and vision with open, customizable models (2024)","year":""},{"key":"B42","article-title":"Bert: pre-training of deep bidirectional transformers for language understanding","author":"Devlin","year":""},{"key":"B43","doi-asserted-by":"crossref","DOI":"10.1109\/ICAICTA49861.2020.9428882","article-title":"Fine-tuning pretrained multilingual bert model for indonesian aspect-based sentiment analysis","author":"Azhar","year":""},{"key":"B44","doi-asserted-by":"publisher","first-page":"453","DOI":"10.28991\/HIJ-2023-04-02-015","article-title":"Bert: a review of applications in sentiment analysis","volume":"4","author":"Sayeed","year":"2023","journal-title":"HighTech Innov J"},{"key":"B45","article-title":"Distilbert, a distilled version of BERT: smaller, faster, cheaper and lighter (2019)","author":"Sanh","year":""},{"key":"B46","doi-asserted-by":"publisher","first-page":"597","DOI":"10.1086\/222355","article-title":"Social behavior as exchange","volume":"63","author":"Homans","year":"1958","journal-title":"AJS"},{"key":"B47","doi-asserted-by":"publisher","first-page":"1015921","DOI":"10.3389\/fpsyg.2022.1015921","article-title":"Social exchange theory: systematic review and future directions","volume":"13","author":"Ahmad","year":"2023","journal-title":"Front Psychol"},{"key":"B48","doi-asserted-by":"publisher","first-page":"172","DOI":"10.1016\/0022-1031(80)90007-4","article-title":"Commitment and satisfaction in romantic associations: a test of the investment model","volume":"16","author":"Rusbult","year":"1980","journal-title":"J Exp Soc Psychol"},{"key":"B49","doi-asserted-by":"publisher","first-page":"153","DOI":"10.1007\/s10896-005-9013-z","article-title":"Using the investment model to understand battered women\u2019s commitment to abusive relationships","volume":"21","author":"Rhatigan","year":"2006","journal-title":"J Fam Violence"},{"key":"B50","first-page":"103","article-title":"Online grooming: moves and strategies","volume":"4","author":"Carmody","year":"2017","journal-title":"Lang Law"},{"key":"B51","doi-asserted-by":"publisher","first-page":"62","DOI":"10.1016\/j.avb.2012.09.003","article-title":"A review of online grooming: characteristics and concerns","volume":"18","author":"Whittle","year":"2013","journal-title":"Aggress Violent Behav"},{"key":"B52","volume-title":"A Typology of Cybersexploitation and Online Grooming Practices","author":"O\u2019Connell","year":"2003"},{"key":"B53","doi-asserted-by":"publisher","first-page":"2811","DOI":"10.3390\/molecules24152811","article-title":"Multi-level comparison of machine learning classifiers and their performance metrics","volume":"24","author":"R\u00e1cz","year":"2019","journal-title":"Molecules"},{"key":"B54","doi-asserted-by":"crossref","DOI":"10.1109\/CSET.2019.8904885","article-title":"Predatory conversation detection","author":"Borj","year":""},{"key":"B55","article-title":"Data from: langdetect (2021)","year":""},{"key":"B56","article-title":"Machine learning to detect online grooming (master\u2019s thesis)","author":"Meyer","year":""},{"key":"B57","doi-asserted-by":"publisher","first-page":"108","DOI":"10.1016\/j.csl.2013.04.007","article-title":"Exploring high-level features for detecting cyberpedophilia","volume":"28","author":"Bogdanova","year":"2014","journal-title":"Comput Speech Lang"},{"key":"B58","article-title":"Data from: Llama\u2013responsible use guide (2024)","year":""}],"container-title":["Frontiers in Pediatrics"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fped.2025.1591828\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,6,19]],"date-time":"2025-06-19T05:29:56Z","timestamp":1750310996000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fped.2025.1591828\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,6,19]]},"references-count":58,"alternative-id":["10.3389\/fped.2025.1591828"],"URL":"https:\/\/doi.org\/10.3389\/fped.2025.1591828","relation":{},"ISSN":["2296-2360"],"issn-type":[{"value":"2296-2360","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,6,19]]},"article-number":"1591828"}}