[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-270936-105":59,"doc-detail-270936-en":131},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":124,"head_meta":126,"extra_data":128,"updated_unix":130},105,"en","i-see-an-instruction-tuned-sop-enhanced-quality-evaluator-for-product-content","I-SEE: An Instruction-tuned, SOP-Enhanced Quality Evaluator for Product Content","","High-quality product content is critical for customer satisfaction and conversions across digital platforms, yet ensuring completeness, accuracy, and alignment with expectations is difficult at scale. Existing content evaluation methods treat information uniformly and require extensive manual prompt engineering to encode domain expertise into LLMs. ISEE introduces a unified framework that (1) automatically identifies customer-impacting features from search behavior, queries, and feedback, (2) uses an instruction-tuned multimodal LLM to follow SOP-aligned operational guidelines, and (3) generalizes zero-shot to new products and SOPs through targeted instruction tuning.",{"@graph":69,"@context":123},[70,84,106],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/i-see-an-instruction-tuned-sop-enhanced-quality-evaluator-for-product-content/270936/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/i-see-an-instruction-tuned-sop-enhanced-quality-evaluator-for-product-content/270936.png","ImageObject",300,407,{"name":92,"@type":93},"Hazel","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-30","2026-09-15",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",15,{"@type":107,"mainEntity":108},"FAQPage",[109,115,119],{"name":110,"@type":111,"acceptedAnswer":112},"What problem does ISEE address in e-commerce product content evaluation?","Question",{"text":113,"@type":114},"ISEE targets the challenge of detecting product content defects at scale, where information is often incomplete, inaccurate, or not aligned with customer expectations. It addresses limitations of workflows that treat all features uniformly and rely heavily on manual prompt design.","Answer",{"name":116,"@type":111,"acceptedAnswer":117},"How does ISEE identify customer-impacting features?",{"text":118,"@type":114},"ISEE uses customer signals such as search query logs, product questions, and customer feedback to synthesize and surface features with higher customer impact. This enables prioritization of high-value fixes rather than equal treatment of all features.",{"name":120,"@type":111,"acceptedAnswer":121},"How does ISEE improve defect detection accuracy and SOP compliance?",{"text":122,"@type":114},"ISEE employs an instruction-tuned multimodal LLM trained to reliably follow structured operational guidelines. It also performs targeted instruction tuning to support robust zero-shot generalization to new product content, features, and SOPs.","https://schema.org",{"og:url":83,"og:type":125,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":127,"canonical":83},"index,follow",{"doc_id":129,"site_id":62},270936,1789458681,{"code":4,"msg":5,"data":132},{"doc_id":129,"user_id":133,"nickname":92,"user_avatar":134,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":52,"language":140,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":67,"update_tm":130,"read_time":144,"preview_page_count":4},137441390410,"https://ap-avatar.wpscdn.com/avatar/2000252f4ab5702993?_k=1776741390130283984","I-SEE: An Instruction-tuned, SOP-Enhanced Quality Evaluator for  \nProduct Content  \nAniket Joshi*  \nAmazon  \n[anikjosh@amazon.com](anikjosh@amazon.com)  \nCyrus DSouza*  \nAmazon  \n[dsocyrus@amazon.com](dsocyrus@amazon.com)  \nSejal Jain  \nAmazon  \n[sejaljn@amazon.com](sejaljn@amazon.com)  \nJitenkumar Rana  \nAmazon  \n[jitenkra@amazon.com](jitenkra@amazon.com)  \nPromod Yenigalla  \nAmazon  \n[promy@amazon.com](promy@amazon.com)  \nAbstract  \nHigh-quality content is critical for driving customer satisfaction and conversions across digital platforms and e-commerce. Ensuring that essential information is complete, accurate, and aligned with customer expectations presentsa significant challenge at scale. Existing approaches to content evaluation often treat all information uniformly, without prioritizing based on customer relevance, and rely heavily on manual prompt design to encode domain expertise into Large Language Models (LLMs) . We present ISEE, a unified framework that addresses these limitations through three core innovations: (1) automated identification of customer-impacting features by synthesizing signals from search behavior, queries, and feedback, enabling targeted content improvements; (2) an instruction-tuned multimodal LLM trained to reliably follow structured operational guidelines, reducing dependence on manual prompt engineering; and (3) robust zeroshot generalization to new product content, features and SOPs via targeted instruction tuning.  \nEvaluated across 20 product categories and 150 product specific features, ISEE achieves 90% precision at 78% recall in detecting product content issues, outperforming much larger (> 200B parameters) models while using a compact 12B architecture.  \n1 Introduction  \nHigh-quality product content is critical for customer trust and conversions in e-commerce. Studies in e-commerce have shown that incomplete or inaccurate information on product pages negatively impacts customer trust and conversions (Amsl et al., 2023)  \nThese issues typically exists in different forms, for example-a conflict between values in different sections (e.g., \"16GB RAM\" in title vs. \"32GB\" in  \n* Equal contribution.  \nfeatures), semantic inaccuracies or misleading description (e.g., labeling a GPU as \"Intel Pentium\") and omission of important information (e.g., cabin bag capacity) .  \nDefect detection in product details requires analyzing multiple data sources, such as search query logs and customer interaction data, customer feedback, product content etc. Today, this process is largely ad-hoc, relying on domain expertise without a systematic approach. Existing workflows often attempt to detect defects across all features or use static \"relevant\" lists, ignoring actual customer importance. This one-size-fits-all strategy leads to wasted effort on low-impact issues.  \nFor example, fixing an issue such as the font style on the keycaps of a keyboard may take priority over more critical problems like missing size specifications (e.g., full-size vs. compact) or layout information (e.g., US vs. UK) for the keyboard, or incorrect RAM information in a gaming laptopissues with real impact on conversions.  \nWhile LLMs are increasingly used for product page evaluation, they still rely on manually authored prompts created by domain experts for each product-feature pair. Moreover, LLMs often struggle to follow detailed Standard Operating Procedures (SOPs) (Hwang et al., 2025 ; Reddyet al., 2023 ; Ouyang et al., 2022), defaulting topre-trained knowledge despite explicit instructions (Figure 1) . As a result, teams spend over 60 expert hours monthly on iterative prompt refinement, complicating the detection of subtle defects.  \nTo address these limitations, we propose ISEE-an SOP-aligned Evaluator for online product details, with the following key innovations:  \nCustomer-aware prioritization: ISEE identifies high-impact features instead of treating all features equally by analyzing search query logs, product questions, and ","cbCaioZltFqBLb8G","https://ap.wps.com/l/cbCaioZltFqBLb8G","pdf",889568,"English","# Introduction\n## Customer-aware prioritization\n## Instruction-aligned defect detection\n## Zero-shot generalization\n# System Overview\n## Production workflow and defect routing","[{\"question\":\"What problem does ISEE address in e-commerce product content evaluation?\",\"answer\":\"ISEE targets the challenge of detecting product content defects at scale, where information is often incomplete, inaccurate, or not aligned with customer expectations. It addresses limitations of workflows that treat all features uniformly and rely heavily on manual prompt design.\"},{\"question\":\"How does ISEE identify customer-impacting features?\",\"answer\":\"ISEE uses customer signals such as search query logs, product questions, and customer feedback to synthesize and surface features with higher customer impact. This enables prioritization of high-value fixes rather than equal treatment of all features.\"},{\"question\":\"How does ISEE improve defect detection accuracy and SOP compliance?\",\"answer\":\"ISEE employs an instruction-tuned multimodal LLM trained to reliably follow structured operational guidelines. It also performs targeted instruction tuning to support robust zero-shot generalization to new product content, features, and SOPs.\"}]","I-SEE: An Instruction-tuned, SOP-Enhanced Quality Evaluator for Product Content | PDF",25]