[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-128533-en":3,"doc-seo-128533-105":30,"detail-sidebar-cat-0-en-105":92},{"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":20,"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},128533,687207020761,"Patrick","https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d",8,"Research & Report","AN ENERGY-BASED COMPARATIVE ANALYSIS OF COMMON APPROACHES TO TEXT CLASSIFICATION IN THE LEGAL DOMAIN","Machine Learning research often prioritizes predictive accuracy, but gaps in performance between text classification methods can be small compared with the impact of production and deployment costs. This study quantitatively compares Large Language Models and traditional approaches such as SVM on the LexGLUE benchmark using both standard metrics (e.g., F1) and green metrics including timing, power consumption, estimated costs, and carbon footprint (CO2). It evaluates prototyping and in-production separately, showing lightweight models can match strong F1 with substantially lower energy demand and resource requirements.","AN ENERGY-BASED COMPARATIVE ANALYSIS OF COMMON APPROACHES TO  \nTEXT CLASSIFICATION IN THE  \nLEGAL DOMAIN  \nSinan Gultekin, Achille Globo, Andrea Zugarini, Marco Ernandes,  \nand Leonardo Rigutini  \nDepartment of Hybrid Linguistic Technologies [expert.ai](expert.ai), Siena, Italy  \nABSTRACT  \nMost Machine Learning research evaluates the best solutions in terms of performance.  \nHowever, in the race for the best performing model, many important aspects are often overlooked when, on the contrary, they should be carefully considered. Infact, sometimes the gaps in performance between different approaches are neglectable, whereas factors such as production costs, energy consumption, and carbon footprint must take into consideration. Large Language Models (LLMs) are extensively adopted to address NLP problems in academia and industry. In this work, we present a detailed quantitative comparison of LLM and traditional approaches (e.g. SVM) on the LexGLUE benchmark, which takes into account both performance (standard indices) and alternative metrics such as timing, power consumption and cost, in a word: the carbon-footprint. In our analysis, we considered the prototyping phase (model selection by training-validation-test iterations) and in-production phases separately, since they follow different implementation proceduresand also require different resources. The results indicate that very often, the simplest algorithms achieve performance very close to that of large LLMs but with very low power consumption and lower resource demands. The results obtained could suggest companies to include additional evaluations in the choice of Machine Learning (ML) solutions.  \nKEYWORDS  \nNLP, text mining, green AI, green NLP, carbon footprint, energy consumption, evaluation.  \n1. INTRODUCTION  \nOver the past decade, we have observed a critical paradigm shift in the field of NLP. The increasing diffusion of end-to-end approaches led to the development of a broad set of Large Language Models (LLMs) based on different neural network architectures and consisting of billions of parameters. Given their huge training and deployment costs, these giant models are typically exclusive to the handful of global companies (i.e., Google, Microsoft) that can sustain such costs. They are typically released as pre-trained models and require a fine-tuning step torefine the model based on the customer’s requirements. However, they require vast amounts of resources to operate in terms of hardware and energy. Most academics, data scientists, or insiders often ignore aspects of energy consumption, but the increasing energy-hungry computation trend raises some relevant concerns. From an ethical and social point of view, we are all witnesses to severe climate change due to pollution and CO2 emissions. From an economic and industrial  \npoint of view, however, in recent years, the energy cost has reached extremely high levels, and having good light Machine Learning solutions can be of vital for companies.  \nIn this article we present a comparative analysis of two widely used families of text classification models in terms of performance and power consumption. In particular, the investigation aims to explore the balance between the performance and carbon footprint of several models based on (1) Large Language Models (LLM) and on (2) Support Vector Machines (SVM) when employed in a vertical domain. On the performance side, the standard classification metric F1 is considered, while on the green side, the energy consumption (KWh), the estimated costs ( ~~ C~~) and CO2 production are valued. The tests were carried out using the LexGLUE benchmark and the results show that, in many cases, lightweight models obtain excellent performance at significantly lower costs. These results suggest further in-depth studies on the use of Deep Learning approaches in industry and underline the need to consider several aspects in addition to the quality of the predictions when selecting the best ML so","cbCaifX6Cjca1sKK","https://ap.wps.com/l/cbCaifX6Cjca1sKK","pdf",776950,1,11,"English","en",105,"# Introduction\n## Performance vs green metrics\n# Related work and motivation\n## Efficiency and energy-aware evaluation\n# Methodology\n## Models and datasets on LexGLUE\n## Prototyping vs in-production phases\n# Results and discussion\n## Emerging considerations and trade-offs\n# Conclusions and future work","[{\"question\":\"What models are compared for legal-domain text classification?\",\"answer\":\"The paper compares Large Language Models with traditional approaches such as Support Vector Machines (SVM), assessed on the LexGLUE benchmark in a vertical legal domain setting.\"},{\"question\":\"Which evaluation metrics are used in the study?\",\"answer\":\"It uses standard classification performance such as F1, and green-oriented metrics including energy consumption (KWh), estimated cost, CO2 production, and also timing-related measures.\"},{\"question\":\"How does the study account for different deployment stages?\",\"answer\":\"It analyzes prototyping separately from in-production, because these phases follow different implementation procedures and require different resources.\"}]","AN ENERGY-BASED COMPARATIVE ANALYSIS OF COMMON APPROACHES TO TEXT CLASSIFICATION IN THE LEGAL DOMAIN | PDF",1786001587,28,{"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":87,"head_meta":89,"extra_data":91,"updated_unix":28},"an-energy-based-comparative-analysis-of-common-approaches-to-text-classification-in-the-legal-domain","",{"@graph":36,"@context":86},[37,54,69],{"@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/an-energy-based-comparative-analysis-of-common-approaches-to-text-classification-in-the-legal-domain/128533/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-23","2026-08-06",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"What models are compared for legal-domain text classification?","Question",{"text":76,"@type":77},"The paper compares Large Language Models with traditional approaches such as Support Vector Machines (SVM), assessed on the LexGLUE benchmark in a vertical legal domain setting.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"Which evaluation metrics are used in the study?",{"text":81,"@type":77},"It uses standard classification performance such as F1, and green-oriented metrics including energy consumption (KWh), estimated cost, CO2 production, and also timing-related measures.",{"name":83,"@type":74,"acceptedAnswer":84},"How does the study account for different deployment stages?",{"text":85,"@type":77},"It analyzes prototyping separately from in-production, because these phases follow different implementation procedures and require different resources.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":93},[94,98,102,106,111,116,121,124,129,132,136],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":46,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":112,"doc_module":4,"doc_module_name":46,"category_name":113,"show_sort_weight":114,"slug":115},6,"Technology",50,"technology",{"id":117,"doc_module":4,"doc_module_name":46,"category_name":118,"show_sort_weight":119,"slug":120},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":122,"slug":123},30,"research-report",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":126,"show_sort_weight":127,"slug":128},9,"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":46,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":133,"doc_module":4,"doc_module_name":46,"category_name":134,"show_sort_weight":133,"slug":135},10,"Lifestyle","lifestyle",{"id":137,"doc_module":4,"doc_module_name":46,"category_name":138,"show_sort_weight":107,"slug":139},19,"General","general"]