🧀 BigCheese.ai

Social

PGVector's Missing Features

🧀

PGVector is identified as lacking some key features for search solutions, including the handling of required and negated words, explainability with highlights, performant filters and order by, support for sparse vectors, BM25, and other advanced search modes. Trieve highlights these downsides, presenting itself as a more adept solution that maintains simplicity while providing both semantic and fulltext search capabilities. The post explains the improvements Trieve offers and advises on using PGVector carefully considering its limitations.

  • PGVector does not support queries with negated words efficiently.
  • PGVector lacks a feature to highlight keywords in search results.
  • PGVector is slow on filter and order by queries.
  • Trieve has secured $3.5M in funding.
  • Trieve offers a search engine that includes features PGVector is missing.