Skip to content
staticseek

Index Types

staticseek offers several index types to optimize search performance based on your specific needs:

  • LinearIndex:

    • Description: The standard implementation and default choice for most applications.
    • Use Case: Best suited for small to medium-sized datasets where quick exact matches are required.
    • Performance: Provides reliable performance but may degrade with larger datasets.

  • GPULinearIndex:

    • Description: A WebGPU-accelerated implementation that enhances search speed for fuzzy queries.
    • Use Case: Ideal for applications with GPU support that require fast fuzzy search capabilities.
    • Performance: Achieves significant speed improvements, especially for larger datasets, while maintaining accuracy.

  • HybridTrieBigramInvertedIndex:

    • Description: A high-performance implementation designed for larger datasets, utilizing a combination of Trie and bigram indexing.
    • Use Case: Best for applications that require fast search performance across multilingual datasets.
    • Performance: Offers substantial speed improvements but may introduce higher false positive rates for languages like Japanese and reduced accuracy in fuzzy searches.

Choosing the Right Index

Section titled “Choosing the Right Index”

When selecting an index type, consider the following factors:

  • Dataset Size: For smaller datasets, LinearIndex may suffice. For larger datasets, consider GPULinearIndex or HybridTrieBigramInvertedIndex.
  • Search Requirements: If fuzzy search is a priority, GPULinearIndex is recommended. For precise searches in CJK languages, HybridTrieBigramInvertedIndex may be more suitable.
  • Hardware Availability: Ensure that your deployment environment supports the necessary hardware for GPU acceleration if choosing GPULinearIndex.

By understanding the strengths and weaknesses of each index type, you can make informed decisions that enhance the search experience for your users.