[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-118959-en":3,"doc-seo-118959-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},118959,4810365810221,"Aurora","https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d",8,"Research & Report","Exploiting Large Language Models to Train Automatic Detectors of Sensitive Data - thesis","This thesis proposes an automated system to identify sensitive data in text documents in line with the definitions and requirements of the General Data Protection Regulation (GDPR). It surveys current approaches to Personally Identifiable Information (PII) and sensitive data detection, and details how Natural Language Processing (NLP) machine learning models are adapted for the task. A major contribution addresses dataset acquisition by using Large Language Models (LLMs) to generate synthetic training datasets. Both proprietary and open-source LLMs are evaluated to study local generation capabilities. A comprehensive detection framework covers six key domains with context-aware criteria, while Transformer-based models such as BERT are assessed using fine-grained metrics, with the NER model reaching the best results (90%) on interchanged datasets.","UNIVERSITY OF PADOVA  \nDEPARTMENT OF MATHEMATICS  \nMaster degree in Data Science  \nExploiting Large Language Models to Train Automatic Detectors of Sensitive Data  \nSupervisor Master Candidate  \nProfessor Alberto Testolin Simone De Renzis  \nStudent ID: 2056102  \nCo-supervisor Dott. Dennis Dosso Siav S.p.A.  \nAcademic Year  \n2022–2023  \nii  \niv  \nAbstract  \nThis thesis proposes an automated system designed to identify sensitive data within text documents, aligning with the definitions and regulations outlined in the General Data Protection Regulation (GDPR) . It reviews the current state of the art in Personally Identifiable Information (PII) and sensitive data detection, and how machine learning models for Natural Language Processing (NLP) are tailored to perform these tasks. A critical challenge addressed in this work pertains to the acquisition of suitable datasets for the training and evaluation of the proposed system. To overcome this obstacle, we explore the use of Large Language Model (LLM)s to generate synthetic datasets, thus serving as a valuable resource for training classification models. Both proprietary and open-source LLMs are leveraged to investigate the capabilities of local models in document generation. It then presents a comprehensive framework for sensitive data detection, covering six key domains and proposing specific criteria to identify the disclosure of sensitive data, which take into account the context and the domain relevance. To achieve the detection of sensitive data, a variety of models are explored, mainly based on the Transformer architecture (Bidirectional Encoder Representations from Transformers (BERT)), adapted to fulfill tasks of text classification and Named Entity Recognition (NER) . It evaluates the performance of the models using fine-grained metrics, and shows that theNER model achieves the best results (90% score) when trained interchangeably on both datasets, also confirming the quality of the dataset generated with the open source LLM.  \nviii  \nContents  \nAbstract vii  \nList of figures xi  \nList of tables xiii  \n1 Introduction 1  \n1.1 Siav S.p.A........................................... 1  \n1.2 Sections overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2  \n2 Related works 5  \n2. 1 Personal data in GDPR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5  \n2.2 Personal Data detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6  \n2.2. 1 PII . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6  \n2.2.2 Sensitive data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7  \n2.3 Models for NLP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8  \n2.3.1 Word embeddings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8  \n2.3.2 Convolutional Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . 9  \n2.3.3 Recurrent Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10  \n2.3.4 Transformers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11  \n2.4 Dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14  \n2.4.1 Synthetic dataset generation ............................. 15  \n2.5 Large Language Models ................................... 16  \n2.5. 1 Generative Pre-trained Transformer (GPT) . . . . . . . . . . . . . . . . . . . . . 16  \n2.5.2 LLaMa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17  \n2.5.3 Generation parameters ................................ 17  \n2.5.4 Prompt Engineering ................................. 18  \n2.5.5 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19  \n3 Methodologies 21  \n3. 1 Sensitive data definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21  \n3.1.1 Topic Clas","cbCaiqSWFmNnWGta","https://ap.wps.com/l/cbCaiqSWFmNnWGta","pdf",17793278,1,85,"English","en",105,"# Abstract\n# List of figures\n# List of tables\n# 1 Introduction\n## 1.1 Siav S.p.A.\n## 1.2 Sections overview\n# 2 Related works\n## 2.1 Personal data in GDPR\n## 2.2 Personal Data detection\n## 2.3 Models for NLP\n## 2.4 Dataset\n## 2.5 Large Language Models\n# 3 Methodologies\n## 3.1 Sensitive data definition\n## 3.2 Implementation code and libraries\n## 3.3 Dataset\n## 3.4 Models\n## 3.5 Evaluation\n# 4 Experiments\n## 4.1 Results","[{\"question\":\"How does the thesis detect sensitive data in text documents?\",\"answer\":\"It proposes a framework that applies context-aware criteria across six key domains and uses Transformer-based models for tasks like text classification and Named Entity Recognition (NER).\"},{\"question\":\"Why are Large Language Models (LLMs) used in the system?\",\"answer\":\"LLMs are used to generate synthetic datasets, addressing the challenge of acquiring suitable datasets for training and evaluation.\"},{\"question\":\"Which model performs best and how is it evaluated?\",\"answer\":\"The NER model achieves the best results (90% score) when trained interchangeably on both datasets, evaluated using fine-grained metrics including binary and multiclass evaluation.\"}]","Exploiting Large Language Models to Train Automatic Detectors of Sensitive Data - thesis | PDF",1785721203,214,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"exploiting-large-language-models-to-train-automatic-detectors-of-sensitive-data-thesis","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/exploiting-large-language-models-to-train-automatic-detectors-of-sensitive-data-thesis/118959/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-03",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"How does the thesis detect sensitive data in text documents?","Question",{"text":75,"@type":76},"It proposes a framework that applies context-aware criteria across six key domains and uses Transformer-based models for tasks like text classification and Named Entity Recognition (NER).","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Why are Large Language Models (LLMs) used in the system?",{"text":80,"@type":76},"LLMs are used to generate synthetic datasets, addressing the challenge of acquiring suitable datasets for training and evaluation.",{"name":82,"@type":73,"acceptedAnswer":83},"Which model performs best and how is it evaluated?",{"text":84,"@type":76},"The NER model achieves the best results (90% score) when trained interchangeably on both datasets, evaluated using fine-grained metrics including binary and multiclass evaluation.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]