[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-123112-en":3,"doc-seo-123112-105":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},123112,687197207639,"Asher","https://ap-avatar.wpscdn.com/davatar_a8503ba1806abce46bf441b54a3ca4cd",8,"Research & Report","Emotion on the edge - An evaluation of feature representations and machine learning models","The study provides a detailed comparison of textual emotion classification methods using a tweet-based dataset labeled with surprise, love, fear, anger, sadness, and joy. Nine machine learning classifiers are tested with Bag of Words and TF-IDF feature representations, alongside a fine-tuned DistilBERT transformer. Training and inference time measurements are conducted on an edge board to evaluate efficiency under constrained computation. Results show feature representations strongly affect effectiveness, with BoW and TF-IDF outperforming DistilBERT. TF-IDF offers better overall performance and faster fitting time, while SGD and SVM lead in inference efficiency.","Emotion on the edge: An evaluation of feature representations and machine learning models  \nJames Thomas Black ∗, Muhammad Zeeshan Shakir  \nSchool of Computing, Engineering and Physical Sciences, University of the West of Scotland, Paisley, PA1 2BE, United Kingdom  \n| A R T I C L E I N F O |  | A B S T R A C T |\n| --- | --- | --- |\n| Keywords:\u003Cbr>Emotion classification Bag of words\u003Cbr>TF-IDF\u003Cbr>Natural language processing DistilBERT |  | This paper presents a comprehensive analysis of textual emotion classification, employing a tweet-based dataset to classify emotions such as surprise, love, fear, anger, sadness, and joy. We compare the performances of nine distinct machine learning classification models using Bag of Words (BoW) and Term Frequency-Inverse Document Frequency (TF-IDF) feature representations, as well as a fine-tuned DistilBERT transformer model. We examine the training and inference times of models to determine the most efficient combination when employing an edge architecture, investigating each model’s performance from training to inference using an edge board. The study underscores the significance of combinations of models and features in machine learning, detailing how these choices affect model performance when low computation power needs to be considered. The findings reveal that feature representations significantly influence model efficacy, with BoW and TF-IDF models outperforming DistilBERT. The results show that while BoW models tend to have higher accuracy, the overall performance of TF-IDF models is superior, requiring less time for fitting, Stochastic Gradient Descent and Support Vector Machines proving to be the most efficient in terms of performance and inference times. |\n\n1. Introduction  \nText classification is a process in the field of Natural Language Processing (NLP) which encompasses a diverse array of specific tasks, including but not limited to sentiment analysis, question classification, and topic classification. It presents a perennial challenge: the selection of optimal models and feature representations that are best suited for each specific task. This paper aims to delve into this challenge by undertaking a comparative analysis of the performance of nine distinct machine learning classification models and a DistilBERT model, utilising three different feature representations.  \nText classification can be executed through various approaches, which range from rule-based systems, Pietramala et al. (2008), through traditional machine learning, Kadhim (2019), to more contemporary deep learning methods, Minaee et al. (2021). Each of these methods has its own merits, offering varying degrees of success. However, deep learning models are often computationally intensive, making them less suitable for edge computing environments where resources are limited. In this work we show that traditional machine learning methods, with their lighter computational footprint, provide a feasible alternative for emotion classification on edge devices with often comparable or improved accuracy. A pivotal factor influencing the performance of machine learning models is feature representation, with traditional methods such as Bag of Words (BoW), Zhang et al. (2010),  \n∗ Corresponding author.  \nE-mail address: [james.black@uws.ac.uk](james.black@uws.ac.uk) (J.T. Black).  \nand Term Frequency-Inverse Document Frequency (TF-IDF), Aizawa (2003), placed against more modern techniques such as word embeddings, Gutiérrez and Keith (2019) or transformer models like Bidirectional Encoder Representations from Transformers (BERT), Devlin et al.(2019).  \nEmotion classification, a subset of text classification, involves the identification of fine-grained emotions such as happiness, sadness, andanger from textual sources, Acheampong et al. (2020). These sources often take informal forms such as social media posts, text messages, or blog content, a trend that is increasingly prevalent in the field.  \nEdge computing has many use ","cbCair36Vf6JJnFZ","https://ap.wps.com/l/cbCair36Vf6JJnFZ","pdf",1166708,1,11,"English","en",105,"# Introduction\n## Background: text classification and feature selection\n## Models and feature representations compared\n## Edge computing context and motivation\n# Methods and experimental setup\n## Dataset and emotion labels\n## Compared classifiers and feature sets\n## Edge deployment and timing evaluation\n# Results and discussion\n## Impact of feature representations\n## Efficiency trade-offs: training vs inference\n## Most efficient model combinations","[{\"question\":\"What emotions and dataset are used for the emotion classification experiments?\",\"answer\":\"The work uses a tweet-based dataset curated by Saravia et al. (2018), labeled into six emotions: surprise, love, fear, anger, sadness, and joy.\"},{\"question\":\"Which feature representations and models are compared in the paper?\",\"answer\":\"Nine machine learning classifiers are evaluated with Bag of Words and TF-IDF feature representations, and a fine-tuned DistilBERT transformer is also included for comparison.\"},{\"question\":\"How does the paper evaluate efficiency for edge deployment?\",\"answer\":\"The study measures both training and inference times and analyzes performance on an edge board to determine efficient combinations under limited computation power.\"}]","Emotion on the edge - An evaluation of feature representations and machine learning models | PDF",1785814704,28,{"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":86,"head_meta":88,"extra_data":90,"updated_unix":28},"emotion-on-the-edge-an-evaluation-of-feature-representations-and-machine-learning-models","",{"@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/emotion-on-the-edge-an-evaluation-of-feature-representations-and-machine-learning-models/123112/",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","2026-08-04",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},"What emotions and dataset are used for the emotion classification experiments?","Question",{"text":75,"@type":76},"The work uses a tweet-based dataset curated by Saravia et al. (2018), labeled into six emotions: surprise, love, fear, anger, sadness, and joy.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Which feature representations and models are compared in the paper?",{"text":80,"@type":76},"Nine machine learning classifiers are evaluated with Bag of Words and TF-IDF feature representations, and a fine-tuned DistilBERT transformer is also included for comparison.",{"name":82,"@type":73,"acceptedAnswer":83},"How does the paper evaluate efficiency for edge deployment?",{"text":84,"@type":76},"The study measures both training and inference times and analyzes performance on an edge board to determine efficient combinations under limited computation power.","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"]