Glossary » search
the mechanism that enables users to find things in large information spaces. Search issues fall into 3 common categories: how people specify their search queries, what determines a “relevant” match, and how search results are displayed.
- query syntax: typical options for specifying a search query include boolean expressions, natural language queries, parametric search, refining previous queries, scoped search, similarity requests (“find documents like this one”), and the typical web query style (keyword with quotes to identify multi-word strings and + and – to indicate required or excluded terms).
- relevance: what determines a good match for a query is a major subject of research. Algorithms rely on frequency of keywords, order, prominence, expert indexing, and popularity of a result for a given keyword. Advanced search engines often consider these types of capabilities — spell-checking, synonym substitution (if they look for ‘jobs’ they might also want ‘careers’), stop words (the term ‘web’ doesn’t usefully distinguish web pages, so ignore it), fuzzy queries (allow non-exact matches), and stem words (if they asked for ‘biking’ also look for ‘bike’).
- displaying results: some issues in displaying results include how to rank results, how to group results (by category, domain name, etc.) and visualize relationships between them, how to display results within a structured document, what information to show about each match, and how many matches to display. Since most searches result in too many or too few results, the interface needs to provide useful search tips suggesting how to reformulate queries for improved results.