How does Filmot's full-text search work, and does it use AI or semantic search?
Filmot independently collects public YouTube video metadata and available subtitles, captions or transcripts, builds a full-text index from that information and searches the stored index.
Subtitle and metadata searches are keyword-based; Filmot does not offer AI-based semantic or contextual search. To refine matches, use spaces to require all terms, | for alternatives, -word to exclude a term, parentheses for groups and double quotes for an exact phrase. Proximity searches with ~N, NEAR/N or NOTNEAR/N can match nearby words, and * inside a quoted phrase stands for one word.
Combine those expressions with filters such as channel, language, upload date or video duration. The ? icon beside the search box opens the Search Syntax Cheat Sheet with examples. Details of Filmot's underlying software and infrastructure are not published.