[[["わかりやすい","easyToUnderstand","thumb-up"],["問題の解決に役立った","solvedMyProblem","thumb-up"],["その他","otherUp","thumb-up"]],[["わかりにくい","hardToUnderstand","thumb-down"],["情報またはサンプルコードが不正確","incorrectInformationOrSampleCode","thumb-down"],["必要な情報 / サンプルがない","missingTheInformationSamplesINeed","thumb-down"],["翻訳に関する問題","translationIssue","thumb-down"],["その他","otherDown","thumb-down"]],["最終更新日 2025-09-02 UTC。"],[],[],null,["# Interactive demo\n\nExperience the power of state-of-the-art vector search technology with the\nVector Search interactive demo. Leveraging real-world datasets,\nthe demo provides a realistic example that will help you learn how\nVector Search works, explore semantic and hybrid search,\nand see reranking in action. Submit a brief description of an animal,\nplant, ecommerce merchandise, or other item, and let Vector Search do\nthe rest!\n| **Note:** Datasets are provided by [Mercari](https://www.mercari.com/) (a popular online marketplace in the US and Japan) and [GBIF (Global Biodiversity Information Facility)](https://www.gbif.org/).\n\n\u003cbr /\u003e\n\nTry it!\n-------\n\nExperiment with the different options in the demo to get a head start with\nVector Search and understand the basics of vector search technology.\n\nTo run:\n\n1. In the **Query** text field, describe the items you want to query for\n (for example, `vintage 1970s pinball machine`). Alternatively, click\n **Generate Query** to auto-generate a description.\n\n2. Click **Submit**.\n\nTo learn more about what you can do in the demo,\nsee [User Interface](/vertex-ai/docs/vector-search/overview#try-ui). \n\n\u003cbr /\u003e\n\nUser Interface\n--------------\n\nThis section describes settings in the UI you can use to control the results\nVector Search returns and how they are ranked.\n\n*** ** * ** ***\n\n### Dataset\n\nUse the **Dataset** drop-down to choose which dataset Vector Search\nwill run your query against. See [Datasets](#datasets) for details\nabout each one.\n\n*** ** * ** ***\n\n### Query\n\nFor the **Query** field, add a description or one or more keywords to specify\nwhat items you want Vector Search to find. Alternatively, click\n**Generate Query** to auto-generate a description.\n\n*** ** * ** ***\n\n### Modify\n\nSeveral options are available that modify the results Vector Search\nreturns:\n\n- Click **Rows** and choose the maximum number of search results that you want\n Vector Search to return.\n\n- Select **Use dense embeddings** if you want Vector Search to return\n semantically similar results.\n\n- Select **Use sparse embeddings** if you want Vector Search to return\n results based on your query's text syntax. Not all available datasets support\n sparse embedding models.\n\n- Select both **Use dense embeddings** and **Use sparse embeddings** if you want\n Vector Search to use hybrid search. Not all datasets support\n this model. Hybrid search combines elements of both dense and sparse embeddings\n which can improve the quality of search results. To learn more, go to\n [About hybrid search](/vertex-ai/docs/vector-search/about-hybrid-search).\n\n- In the **RRF Alpha** field, enter between 0.0 and 1.0 to specify RRF ranking\n effect.\n\n- To rerank search results, select **ranking_api** from the **Reranking**\n drop-down or select **None** to disable reranking.\n\n*** ** * ** ***\n\n### Metrics\n\nAfter a query runs, you are provided with latency metrics that breakdown the\ntime it took for different stages of search to complete.\n\n*** ** * ** ***\n\n### Query Process\n\nWhen a query is processed, the following occurs:\n\n1. **Query embedding generation:** An embedding is generated for the specified\n query text.\n\n2. **Vector Search query:** The query is run with the Vector Search\n index.\n\n3. **Vertex AI Feature Store fetch:** Features are read (for example, item name, description,\n or image URL) from Vertex AI Feature Store using the list of\n item IDs Vector Search returns.\n\n4. **Reranking:** Retrieved items are sorted through ranking APIs which uses\n query text, item name, and item description to calculate relevance score.\n\n### Embeddings\n\n**Multimodal:** Multimodal semantic search on item images. For details, go to\n[What is Multimodal Search: \"LLMs with vision\" change businesses](https://cloud.google.com/blog/products/ai-machine-learning/multimodal-generative-ai-search).\n\n**Text (semantic similarity):** Text semantic search on item names and\ndescriptions based on semantic similarity. To learn more, go to\n[Vertex AI Embeddings for Text: Grounding LLMs made easy](https://cloud.google.com/blog/products/ai-machine-learning/how-to-use-grounding-for-your-llms-with-text-embeddings).\n\n**Text (question-answering):** Text semantic search on item names and descriptions,\nwith improved search quality by task type QUESTION_ANSWERING. This is suited for\nQ\\&A types of applications. For information about task type embeddings, go to\n[Enhancing your gen AI use case with Vertex AI embeddings and task types](https://cloud.google.com/blog/products/ai-machine-learning/improve-gen-ai-search-with-vertex-ai-embeddings-and-task-types).\n\n**Sparse (Hybrid Search):** Keyword (token-based) search on item names and\ndescriptions, generated with the TF-IDF algorithm. For more information, go to\n[About hybrid search](/vertex-ai/docs/vector-search/about-hybrid-search).\n\nDatasets\n--------\n\n| **Note:** Datasets are provided by [Mercari](https://www.mercari.com/) (a popular online marketplace in the US and Japan) and [GBIF (Global Biodiversity Information Facility)](https://www.gbif.org/).\n\nThe interactive demo includes several datasets you can run queries on. Datasets\ndiffer from each other by embedding model, support for sparse embeddings,\nembedding dimensions, and number of stored items.\n\nNext steps\n----------\n\nNow that you've familiar with the demo, you are ready to take a deeper dive into learning how to use Vector Search.\n\n- **[Quickstart](/vertex-ai/docs/vector-search/quickstart):**\n Use a example dataset to create and deploy an index in 30 minutes or less.\n\n- **[Before you begin](/vertex-ai/docs/vector-search/setup/setup):**\n Discover what to do to prepare embeddings and decide the kind of endpoint to\n deploy your index to."]]
