[[["容易理解","easyToUnderstand","thumb-up"],["確實解決了我的問題","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["難以理解","hardToUnderstand","thumb-down"],["資訊或程式碼範例有誤","incorrectInformationOrSampleCode","thumb-down"],["缺少我需要的資訊/範例","missingTheInformationSamplesINeed","thumb-down"],["翻譯問題","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["上次更新時間：2025-09-05 (世界標準時間)。"],[[["\u003cp\u003eThis page outlines the four available recommendation types for media apps: Others You May Like, Recommended for You, More Like This, and Most Popular.\u003c/p\u003e\n"],["\u003cp\u003eEach recommendation type has a default optimization objective, which is typically click-through rate (CTR), but can be customized to conversion rate (CVR) or watch duration per session in certain cases.\u003c/p\u003e\n"],["\u003cp\u003eThe "Recommended for You" feature is often used on home pages and can also be employed on category pages by utilizing filter tags.\u003c/p\u003e\n"],["\u003cp\u003eThe "More Like This" type suggests media based on similarities to a context item, taking into account factors like overlapping categories among the media.\u003c/p\u003e\n"],["\u003cp\u003eThe "Most Popular" recommendation type highlights the media most viewed across all users in recent history, with the option to adjust the time window.\u003c/p\u003e\n"]]],[],null,["# About media app recommendations types\n\nThis page describes the recommendations types that are available when you\n[create a media recommendations app](/generative-ai-app-builder/docs/create-app-media),\nincluding the business objectives, context event types, and other customizations\nthat are available with each recommendations type.\n\nRecommendations types\n---------------------\n\nThe following recommendations types are available with media recommendations\napps:\n\n- [Others You May Like](#others-you-may-like)\n- [Recommended for You](#recommended-for-you)\n- [More Like This](#more-like-this)\n- [Most Popular](#most-popular)\n\n### Others You May Like\n\nThe Others You May Like recommendations type recommends the next document that a\nuser is most likely to engage with. The recommendation is based on the user's\nengagement history and the current context document.\n\n**Default optimization objective:** [click-through rate](#ctr)\n\n**Default serving config:** N/A\n\n**Available customizations:**\n\n- Business objective\n\n**Supported pages for app deployment:**\n\n- Detail page. For more information, see the [`view-item`](/generative-ai-app-builder/docs/media-user-events#view-item) and [`media-play`](/generative-ai-app-builder/docs/media-user-events#media-play) events.\n\n**User event requirements:**\n\n- `view-item` or `media-play` for click-through rate objective\n\n- `media-complete` and either `media-play` or `view-item` for conversion rate\n objective and watch duration per session objective\n\n### Recommended for You\n\nThe Recommended for You recommendations type predicts the next document that a\nuser is most likely to engage with based on the engagement history of that user.\nThis recommendations type is typically used on the home page.\n\nIf you use Recommended for You on the home page, select Home Page Context as the\ncontext event type for the app. Home Page Context can deliver better performance\nbecause it uses additional usage signals that are not available with the Generic\noption. Only use the Generic option if you don't have `view-home-page` events.\n\n**Default optimization objective:** [click-through rate](#ctr)\n\n**Default context event type:** Home page context\n\n**Default serving config:** N/A\n\n**Available customizations:**\n\n- Business objective\n\n- Context event type\n\n**Supported pages for app deployment:**\n\n- All. On category pages, you must provide filter tags.\n\n**User event requirements:**\n\n- `view-item` or `media-play` for click-through rate objective. For home page\n context, `view-home-page` is also required.\n\n- (`media-play` or `view-item`) and `media-complete` for conversion rate\n objective and watch duration per session objective. For home page context,\n `view-home-page` is also required.\n\n### More Like This\n\nThe More Like This recommendations type recommends media that is similar to a\ncontext item and is likely to be engaged with next by a viewer of the context\nitem. More Like This is based on the context item and the aggregate viewing\nhistory of all users who viewed the context item. A More Like This app is\ntypically used on detail pages or on the home page with a fixed context item.\n\nThe More Like This recommendations type uses a variety of factors to determine\nhow similar two media documents are, including the [categories](/generative-ai-app-builder/docs/media-documents#key-properties) field of the\nmedia documents. For best results, media documents that are similar should have\noverlapping categories---for example, `[\"Action\", \"Comedy\"]` categories are\nsomewhat similar to `[\"Action\", \"Thriller\"]`, but not similar to `[\"Drama\"]`.\n\n**Default optimization objective:** [click-through rate](#ctr)\n\n**Default serving config:** N/A\n\n**Available customizations:**\n\n- Business objective\n\n**Supported pages for app deployment:**\n\n- Detail page\n\n- Home page (requires a context item)\n\n**User event requirements:**\n\n- `view-item` or `media-play` for click-through rate objective\n\n- (`media-play` or `view-item`) and `media-complete` for conversion rate\n objective and watch duration per session objective\n\n### Most Popular\n\nThe Most Popular recommendations type recommends media that has been most\npopular among all users in recent days. The recommendation is based on the\nwatching or viewing history of all users. You can customize the time window to\ncheck the popularity of the documents.\n\n**Default optimization objective:** [click-through rate](#ctr)\n\n**Default serving config:** N/A\n\n**Available customizations:**\n\n- Business objective\n\n- Time window. Specify the time window in days to check the popularity of\n documents in the last *X* days.\n\n**Supported pages for app deployment:**\n\n- Home page\n\n**User event requirements:**\n\n- `view-item` or `media-play` for click-through rate objective\n\n- `media-complete` for conversion rate objective\n\n**Limitations:**\n\n- Most Popular doesn't support customizing serving configs or creating multiple\n serving configs\n\n- Most Popular doesn't support filtering based on categories\n\nOptimization for business objectives\n------------------------------------\n\nThe machine learning models that underlie media recommendation apps are created\nto optimize for a particular business objective, which determines how the model\nis built.\n\nAfter you have trained an app (which is training the underlying model), you\ncannot change the optimization objective. You must train a new app to use a\ndifferent optimization objective.\n\nRecommendations for Media supports the following optimization objectives.\n\n### Click-through rate (CTR)\n\nOptimizing for CTR emphasizes engagement. You should optimize for CTR when you\nwant to maximize the likelihood that the user interacts with the recommendation.\n\nCTR is the default optimization objective for the [Others You May Like](#others-you-may-like)\nand [Recommended for You](#recommended-for-you) recommendations types.\n\n### Conversion rate (CVR)\n\nOptimizing for conversion rate maximizes the likelihood that the user consumes\nthe content up to the conversion threshold defined in the app.\n\nThe conversion threshold can be specified in seconds or percentage. For example,\nif the conversion threshold is set to 25% and the user watches at least 25% of\nthe program, then the conversion objective is met.\n\n### Watch duration per session\n\nOptimizing for watch duration per session maximizes the duration of media\nconsumption. This objective uses information from clicks, conversions, and watch\nduration derived from `media-complete` user events in order to recommend\nitems that have a higher probability of being watched longer than others.\n\nWhen logging user events, make sure that the `mediaProgressDuration` is\nnon-negative and is correctly logged in `media-complete` events in the last 90\ndays.\n\nThe watch duration per session objective is available for [Others You May\nLike](#others-you-may-like), [Recommended for You](#recommended-for-you), and\n[More Like This](#more-like-this) recommendation types.\n\nWhat's next\n-----------\n\n- [Create a media app and a data store](/generative-ai-app-builder/docs/create-app-media)"]]
