Hybrid Search

A sign-language dictionary search pattern that combines search-by-video (a user performs the sign into a camera for sign recognition to match) with search-by-feature (manual filtering of the candidate list by linguistic properties such as handshape and location). Introduced as a strategy for making the most of imperfect sign-recognition AI: the video step removes the burden of specifying an exact feature query, while the filter step lets users narrow a long result list without waiting for recognition accuracy to improve. Hybrid search has been shown to significantly outperform pure search-by-video on user satisfaction, result usefulness, sense of control, and ranking satisfaction — even when the underlying recognition accuracy is identical. The pattern generalises to other human-movement search tasks and to orthographically deep spoken-language lookup.