[[["易于理解","easyToUnderstand","thumb-up"],["解决了我的问题","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["很难理解","hardToUnderstand","thumb-down"],["信息或示例代码不正确","incorrectInformationOrSampleCode","thumb-down"],["没有我需要的信息/示例","missingTheInformationSamplesINeed","thumb-down"],["翻译问题","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["最后更新时间 (UTC)：2025-09-04。"],[[["\u003cp\u003eBigQuery ML supports AI features via Cloud AI APIs, enabling task-specific solutions like natural language processing, machine translation, audio transcription, document processing, and computer vision.\u003c/p\u003e\n"],["\u003cp\u003eYou can access Cloud AI API functionalities by creating a remote model in BigQuery ML, which represents the API endpoint, and then use BigQuery ML functions against the remote model.\u003c/p\u003e\n"],["\u003cp\u003eThis system allows you to utilize the capabilities of the underlying AI APIs without requiring expertise in Python or detailed API familiarity.\u003c/p\u003e\n"],["\u003cp\u003eRemote models over Vertex AI and Cloud AI services can be combined with BigQuery ML functions to perform complex data analysis and generative AI tasks.\u003c/p\u003e\n"],["\u003cp\u003eThe inference for each of the processes (natural language processing, machine translation, audio transcription, document processing and computer vision) occurs in Vertex AI and the results are stored in BigQuery.\u003c/p\u003e\n"]]],[],null,["# Task-specific solutions overview\n================================\n\nThis document describes the artificial intelligence (AI) features that\nBigQuery ML supports. These features let you develop task-specific\nsolutions in BigQuery ML by using Cloud AI APIs. Supported tasks\ninclude the following:\n\n- [Natural language processing](#natural_language_processing)\n- [Machine translation](#machine_translation)\n- [Audio transcription](#audio_transcription)\n- [Document processing](#document_processing)\n- [Computer vision](#computer_vision)\n\nYou access a Cloud AI API to perform one of these functions by creating a\n[remote model](/bigquery/docs/reference/standard-sql/bigqueryml-syntax-create-remote-model-service)\nin BigQuery ML that represents the API endpoint. Once you have\ncreated a remote model over the AI resource that you want to use, you access\nthat resource's capabilities by running a BigQuery ML function\nagainst the remote model.\n\nThis approach lets you use the capabilities of the underlying API\nwithout having to know Python or develop familiarity with API.\n\nWorkflow\n--------\n\nYou can use\n\n[remote models over Vertex AI models](/bigquery/docs/generative-ai-overview)\n\nand\n\nremote models over Cloud AI services\n\ntogether with BigQuery ML functions in order to accomplish\ncomplex data analysis and generative AI tasks.\n\nThe following diagram shows some typical workflows where you might use these\ncapabilities together:\n\nNatural language processing\n---------------------------\n\nYou can use natural language processing to perform tasks such as classification\nand sentiment analysis on your data. For example, you could analyze product\nfeedback to estimate whether customers like a particular product.\n\nTo perform natural language tasks, you can create a reference to the\n[Cloud Natural Language API](/natural-language) by creating a remote model and specifying\n`CLOUD_AI_NATURAL_LANGUAGE_V1` for the `REMOTE_SERVICE_TYPE` value. You can then use the\n[`ML.UNDERSTAND_TEXT` function](/bigquery/docs/reference/standard-sql/bigqueryml-syntax-understand-text)\nto interact with that service. `ML.UNDERSTAND_TEXT` works with data in\n[standard tables](/bigquery/docs/tables-intro#standard_tables). All inference\noccurs in Vertex AI. The results are stored in\nBigQuery.\n\nTo learn more, try\n[understanding text with the `ML.UNDERSTAND_TEXT` function](/bigquery/docs/understand-text).\n\nMachine translation\n-------------------\n\nYou can use machine translation to translate text data into other languages.\nFor example, translating customer feedback from an unfamiliar language into\na familiar one.\n\nTo perform machine translation tasks, you can create a reference to the\n[Cloud Translation API](/translate) by creating a remote model and specifying\n`CLOUD_AI_TRANSLATE_V3` for the `REMOTE_SERVICE_TYPE` value. You can then use the\n[`ML.TRANSLATE` function](/bigquery/docs/reference/standard-sql/bigqueryml-syntax-translate)\nto interact with that service. `ML.TRANSLATE` works with data in\n[standard tables](/bigquery/docs/tables-intro#standard_tables). All inference\noccurs in Vertex AI. The results are stored in\nBigQuery.\n\nTo learn more, try\n[translating text with the `ML.TRANSLATE` function](/bigquery/docs/translate-text).\n\nAudio transcription\n-------------------\n\nYou can use audio transcription to transcribe audio files into written text.\nFor example, transcribing a voicemail recording into a text message.\n\nTo perform audio transcription tasks, you can create a reference to the\n[Speech-to-Text API](/speech-to-text) by creating a remote model and specifying\n`CLOUD_AI_SPEECH_TO_TEXT_V2` for the `REMOTE_SERVICE_TYPE` value. You can\noptionally [specify a recognizer to use](/bigquery/docs/reference/standard-sql/bigqueryml-syntax-create-remote-model-service#speech_recognizer) to process the audio\ncontent. You can then use the\n[`ML.TRANSCRIBE` function](/bigquery/docs/reference/standard-sql/bigqueryml-syntax-transcribe)\nto transcribe audio files. `ML.TRANSCRIBE` works with audio files in\n[object tables](/bigquery/docs/object-table-introduction). All inference\noccurs in Vertex AI. The results are stored in\nBigQuery.\n\nTo learn more, try\n[transcribing audio files with the `ML.TRANSCRIBE` function](/bigquery/docs/transcribe).\n\nDocument processing\n-------------------\n\nYou can use document processing to extract insights from unstructured documents.\nFor example, extracting relevant information from invoice files so it can\nbe input into accounting software.\n\nTo perform document processing tasks, you can create a reference to the\n[Document AI API](/document-ai) by creating a remote model,\nspecifying `CLOUD_AI_DOCUMENT_V1` for the `REMOTE_SERVICE_TYPE` value, and\n[specifying a processor to use](/bigquery/docs/reference/standard-sql/bigqueryml-syntax-create-remote-model-service#document_processor) to process the\ndocument content. You can then use the\n[`ML.PROCESS_DOCUMENT` function](/bigquery/docs/reference/standard-sql/bigqueryml-syntax-process-document)\nto process documents. `ML.PROCESS_DOCUMENT` works on documents in\n[object tables](/bigquery/docs/object-table-introduction). All inference\noccurs in Vertex AI. The results are stored in\nBigQuery.\n\nTo learn more, try\n[processing documents with the `ML.PROCESS_DOCUMENT` function](/bigquery/docs/process-document).\n\nComputer vision\n---------------\n\nYou can use computer vision to perform image analysis tasks. For example, you\ncould analyze images to detect whether they contain faces, or to generate\nlabels describing the objects in the image.\n\nTo perform computer vision tasks, you can create a reference to the\n[Cloud Vision API](/vision) by creating a remote model and\nspecifying `CLOUD_AI_VISION_V1` for the `REMOTE_SERVICE_TYPE` value. You can then use the\n[`ML.ANNOTATE_IMAGE` function](/bigquery/docs/reference/standard-sql/bigqueryml-syntax-annotate-image)\nto annotate images by using that service. `ML.ANNOTATE_IMAGE` works with data in\n[object tables](/bigquery/docs/object-table-introduction). All inference\noccurs in Vertex AI. The results are stored in\nBigQuery.\n\nTo learn more, try\n[annotating object table images with the `ML.ANNOTATE_IMAGE` function](/bigquery/docs/annotate-image).\n\nWhat's next\n-----------\n\n- For more information about performing inference over machine learning models, see [Model inference overview](/bigquery/docs/inference-overview)."]]
