OpenSearch:Select an OpenSearch edition

Edition

Target scenario

Advantages

LLM Intelligent Q&A Edition

Conversational search scenarios such as intelligent customer service, enterprise knowledge bases, content, e-commerce, and education.

• Enterprise-specific Q&A search: Generates Q&A results based on business data for enhanced security and stability.

• Built-in LLM model: No need to access external LLM models. The full search and Q&A pipeline is built in.

• Rich NLP models: Built-in NLP algorithm models including document chunking, text vectorization, and image vectorization.

• Multimodal search: Supports multimodal Q&A result display including text, links, and images to meet personalized Q&A search needs.

• One-stop quick integration: Import your business data to quickly build an enterprise-specific Q&A search service in one stop.

• Zero deployment, fully managed, maintenance-free: A fully managed service that requires no deployment or maintenance, reducing development and O&M costs.

Industry Algorithm Edition

For enterprises and developers who have higher requirements for search quality, want to quickly build an industry-specific intelligent search system in one stop, and need industry best practices, industry query analysis algorithms, industry ranking models, and custom-tuned models. Currently supported industries: e-commerce and retail, gaming, content communities, and education.

• Includes intelligent search algorithm models for industries such as e-commerce, content, and gaming, providing industry search best practices.

• Supports lightweight customization of proprietary tokenizers and NLP algorithm models based on business requirements and characteristics.

• Provides rich search guidance services including auto-suggest, trending searches, and placeholder text, with support for manual operational intervention.

• Out-of-the-box, maintenance-free, with one-click configuration of the entire industry intelligent search pipeline.

High-performance Search Edition

For enterprises and developers who need high-performance big data search for scenarios such as orders, coupons, logistics, and insurance policies, and want to quickly build a search system at low cost in one stop.

• Achieves millisecond-level responses for hundreds of billions of data records based on a proprietary search engine, providing search solutions for scenarios such as orders, logistics, and insurance policies.

• Supports one-click integration with Alibaba Cloud services such as RDS, MaxCompute, and PolarDB, and supports API/SDK access to self-managed databases.

• Provides multiple query methods, supports combined queries, and includes powerful full-text search and prefix/suffix/infix fuzzy queries.

• Supports one-click scaling without worrying about the underlying cluster, with no O&M costs, enabling flexible responses to business scenarios.

Vector Search Edition

For enterprises and developers who need online vector search for scenarios such as image/text search, audio/video search, NLP, Q&A, recommendations, and risk monitoring, and who care about performance, effectiveness, and stability.

• Deeply optimized for vector search scenarios, supporting image search, audio/video search, personalized search and recommendations, Q&A, and other scenarios.

• The minimum specification supports 1 billion 128-dimensional vectors, or thousands of QPS with millisecond-level query responses.

• Built-in multiple vector search algorithms, supporting primary key search, vector search, and namespace-based vector search.

• Adopts a highly available engineering architecture with horizontally scalable data, ensuring data real-time performance and high-concurrency search reliability.

Retrieval Engine Edition

For enterprises and developers who have large-scale text retrieval scenarios with high-performance and low-cost requirements, and need flexible customization of search services.

• Built-in flexible indexing and search strategies, supporting high-performance search for massive data scenarios.

• Scalable to support hundreds of billions of data records, with millisecond-level query responses and millisecond-level data updates.

• Built-in multiple vector search algorithms, supporting hybrid search with tags, text inverted indexes, and vector indexes.

• Supports rich SQL custom functions, engine plugins, and algorithm models for flexible customization of business processing logic.

• Supports standardized SQL development language and IDE to reduce engine learning and development costs.

Feature/Edition

Industry Algorithm Edition

High-performance Search Edition

Retrieval Engine Edition

Vector Search Edition

LLM Intelligent Q&A Edition

Data source

RDS/MaxCompute/PolarDB/API

RDS/MaxCompute/PolarDB/API

OSS/MaxCompute/API

OSS/MaxCompute/API

MaxCompute/API

Table structure

Multi-table offline left join

Multi-table online join (dedicated storage - cluster)

Multi-table online join

Multi-table independent query

Built-in fixed table structure / custom table structure with multi-table offline join

Data update

API/SDK update, and automatic sync update via DTS subscription

API/SDK update, or automatic sync update via DTS subscription

API/SDK update

API/SDK update

API/SDK update

Data backup

When an application has no data source, it supports inheriting data from the online application

Not supported

For API data sources, supports data recovery from historical indexes

For API data sources, supports data recovery from historical indexes

Not supported

Index type

Text, vector

Text

Text, vector

Vector

Text, vector

Analyzer

Industry analyzer, multi-language analyzer, general analyzer, lightweight custom analyzer

Multi-language analyzer, general analyzer

Industry analyzer, multi-language analyzer, general analyzer

Primary key, vector index

Text and vector analyzer

Tokenization intervention

Supported

Supported

Supported

Not supported

Not supported

Query semantic understanding

Stop words, spelling correction, term weighting, synonyms, entity recognition, custom query analysis

Not supported

Not supported

Not supported

Not supported

Ranking method

Ranking expression, ranking plugin (cava script), ranking algorithm (CTR model, category prediction)

Ranking expression, ranking plugin (cava script)

Ranking expression

Ranked by vector similarity score

Ranking expression

Search guidance: auto-suggest, trending searches and placeholder text

Supported

Not supported

Not supported

Not supported

Not supported

SQL syntax

Not supported

Supported (dedicated storage - cluster)

Supported

Not supported

Not supported

A/B Test

Supported

Not supported

Not supported

Not supported

Not supported

Business reports

Supported

Supported

Not supported

Not supported

Not supported

Monitoring and alerting

Supported

Supported

Supported

Supported

Not supported

Slow query optimization

Supported

Supported

Not supported

Not supported

Not supported

Automation level

Maintenance-free

Maintenance-free

Self-service O&M based on system-provided monitoring metrics

Self-service O&M based on system-provided monitoring metrics

Maintenance-free

Region

Industry Algorithm Edition

High-performance Search Edition

Vector Search Edition

Retrieval Engine Edition

LLM Intelligent Q&A Edition

China (Hangzhou)

√

√

√

√

China (Shanghai)

√

√

√

√

√

China (Qingdao)

√

√

China (Beijing)

√

√

√

√

China (Zhangjiakou)

√

√

√

√

China (Shenzhen)

√

√

√

√

China (Hong Kong)

√

Singapore

√

√

√

√

√

Germany (Frankfurt)

√

√

√

US (Virginia)

√

√