[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-661202-105":59,"doc-detail-661202-en":130},{"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":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","information-retrieval-and-retrieval-augmented-generation-chapter-11-intro","Information Retrieval and Retrieval-Augmented Generation - Chapter 11 - Intro","","Chapter 11 introduces how language models satisfy information needs through retrieval. It reviews earlier question-answering systems and motivates large language models for factual answering, then explains why simple prompting fails: hallucinations, poor calibration, and difficulty using proprietary or rapidly changing information stored outside the model. The chapter frames retrieval-augmented generation (RAG) as a solution that grounds generation in external sources such as medical, legal, personal, and corporate documents.",{"@graph":69,"@context":122},[70,84,105],{"@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/information-retrieval-and-retrieval-augmented-generation-chapter-11-intro/661202/",{"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/information-retrieval-and-retrieval-augmented-generation-chapter-11-intro/661202.png","ImageObject",300,407,{"name":92,"@type":93},"Eliana","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-30","2026-10-28",true,{"@type":102,"interactionType":103,"userInteractionCount":19},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"Why are people interested in asking computers questions for information needs?","Question",{"text":112,"@type":113},"The chapter explains that as soon as computers existed, people wanted them to answer questions, including factoid questions expressed in short text.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"What are the main limitations of answering factual questions by simply prompting an LLM?",{"text":117,"@type":113},"It states that LLMs can hallucinate, are often not well-calibrated, and cannot directly use proprietary data such as personal emails or internal documents.",{"name":119,"@type":110,"acceptedAnswer":120},"How does retrieval-augmented generation (RAG) address these limitations?",{"text":121,"@type":113},"RAG provides external knowledge sources (e.g., medical or legal records and corporate documents) to the model so answers are grounded in the documents rather than only in pretrained parameters.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},661202,1793159584,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":19,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":139,"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":129,"read_time":144},4398048949847,"https://ap-avatar.wpscdn.com/avatar/400002536579ef2da7f?_k=1778318612642679267","Speech and Language Processing . Daniel Jurafsky & James H . Martin . Copyright © 2026 . All rights reserved . Draft of August 19, 2026 .  \nCHAPTER  \n11  \nInformation Retrieval and  \nRetrieval-Augmented Generation  \nOn two occasions I have been asked,—“Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?”... I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a  \nquestion. Babbage (1864)  \nPeople need to know things. So pretty much as soon as there were computers we were asking them questions. By 1961 there was a system to answer questions about American baseball statistics like “How many games did the Yankees play in July?” (Green et al., 1961) . Even fictional computers in the 1970s like Deep Thought, invented by Douglas Adams in The Hitchhiker’s Guide to the Galaxy, answered “the Ultimate Question Of Life, The Universe, and Everything”.1 And because so much knowledge is encoded in text, systems were answering questions at human-level performance even before LLMs: IBM’s Watson system won the TV game-show Jeopardy! in 2011, surpassing humans at answering questions like:  \nWILLIAM WILKINSON’S “AN ACCOUNT OF THE  \nPRINCIPALITIES OF WALLACHIA AND MOLDOVIA”  \nINSPIRED THIS AUTHOR’S MOST FAMOUS NOVEL 2  \nIt follows naturally, then, that an important function of large language models is to fill human information needs. And since a lot of information is online, finding the information that fills our needs is closely related to web information retrieval, the task performed by search engines. Indeed, the distinction is becoming ever more fuzzy, as modern search engines are integrated with large language models.  \nqufacesttiooidns Consider some simple information needs, for example factoid questions that can be met with facts expressed in short texts like the following:  \n(11 . 1) Where is the Louvre Museum located?  \n(11.2) Where does the energy in a nuclear explosion come from?  \n(11.3) How to get a script l in latex?  \nTo get an LLM to answer these questions, we can just prompt it! For example apretrained LLM that has been instruction-tuned on answering questions (instructiontuning is in Chapter 8) could directly answer the following question  \nWhere is the Louvre Museum located?  \nby performing conditional generation given this prefix, and take the response as the answer. This works because large language models have processed a lot of facts in their pretraining data, including the location of the Louvre, and have encoded this information in their parameters. Factual knowledge of this type seems to be stored in the connections in the very large feedforward layers of transformer models (Geva et al., 2021 ; Meng et al., 2022) .  \n1 The answer was 42, but unfortunately the question was never revealed.  \n2 The answer, of course, is ‘Who is Bram Stoker’, and the novel was Dracula.  \n2 CHAPTER 11 • RETRIEVAL-BASED MODELS  \nhallucinate  \ncalibrated  \nRAG information retrieval  \nSimply prompting an LLM is useful for many generation tasks, including those involving facts. But the fact that knowledge is stored in the feedforward weights of the LLM leads to a number of problems with prompting as a method for correctly generating factual texts or answers.  \nThe first and main problem is that LLMs are often incorrect when generating answers or other texts about facts! Large language models hallucinate. A hallucination is a response that is not faithful to the facts of the world. That is, when asked questions, large language models sometimes make up answers that sound reasonable. For example, Dahl et al. (2024) found that when asked questions about the legal domain (like about particular legal cases), large language models hallucinated from 69% to 88% of the time! LLMs sometimes give incorrect factual responseseven when the correct facts are stored in the parameters; this seems to be caused by the feedforward layers failing to recall the knowledge stored in their par","cbCaiajNvW13JCRe","https://ap.wps.com/l/cbCaiajNvW13JCRe","pdf",569080,25,"English","# Information Retrieval and Retrieval-Augmented Generation\n## Why LLMs Need Retrieval\n## Problems with Simple Prompting\n## Hallucinations and Calibration\n## Proprietary and Time-Sensitive Knowledge\n## Retrieval-Augmented Generation (RAG)","[{\"question\":\"Why are people interested in asking computers questions for information needs?\",\"answer\":\"The chapter explains that as soon as computers existed, people wanted them to answer questions, including factoid questions expressed in short text.\"},{\"question\":\"What are the main limitations of answering factual questions by simply prompting an LLM?\",\"answer\":\"It states that LLMs can hallucinate, are often not well-calibrated, and cannot directly use proprietary data such as personal emails or internal documents.\"},{\"question\":\"How does retrieval-augmented generation (RAG) address these limitations?\",\"answer\":\"RAG provides external knowledge sources (e.g., medical or legal records and corporate documents) to the model so answers are grounded in the documents rather than only in pretrained parameters.\"}]","Information Retrieval and Retrieval-Augmented Generation - Chapter 11 - Intro | PDF",63]