I've noticed a following asymmetry in search interfaces. While Google/Bing/Yahoo uses quite complicated rules for
- deciding if a document matches a query,
- sorting results
the situation is completely different in a browser or text editor when a user presses CTRL+F :
- a match is simply based on a string matching (no stemming, no semantic, no grammar, no proximity, ...)
- sorting is simply done according to position of match inside the document (not relevance, ...)
Is this asymmetry a reflection of actual user needs in these two scenarios or rather a consequence of less computing power available client-side than server-side?
Context: I'm trying to design rules for search (matching and ordering of results) within a single very long text. The text is a log of chatroom between ~5 users. The metadata includes timestamps, and usernames for each line of text. The users are professionals, who either participated in the conversation, or not, but understand the topic and language of participants. The documents are not publicly available.