Bookmarks

Searchers are constantly challenging search engines with the Turing test, expecting humanlike answers to their sometimes ill-formed questions. Most search engines do well for short keyword queries (like “mars exploration”), but they tend to fail for natural language with syntactic structure (like “space missions before apollo”). Careful search interface design may provide parts of the solution. Query Suggestions help users formulate longer, well-phrased questions, providing optimal answers at the cost of less variety. Best Bets help search engines provide good answers when organic search results fail. Will these two techniques together give humanlike intelligence to the search machine?

With Vertical Lens, Yahoo lets you build your own vertical search engine. You get to mix and rerank your own site content with Yahoo search results, giving searchers relevant results both on and off site. Site content is indexed in real time, and content structure may be used by searchers to navigate and refine their searches. Ranking can be based on freshness, popularity or social factors. Yahoo Vertical Lens is a competitor of Google Site Search and Microsoft Live Search API’s.

Research produced by Yahoo examines consumer behaviour in choosing local services and providers in five vertical categories: legal, real estate, home improvement, health care and vocational education. Did you know that 59 percent of homeowners use the Internet to do research on home improvement contractors, searching mainly for type of service needed and location (city, zip code)? As a local business and service provider, you need to have a professional website that is easily found by consumers using search engines. Local search providers must provide completeness, and novel features based on vertical metadata, in order to distinguish themselves from the majors like Google and Yahoo.

Why is it harder to describe the details of an object, when it's so easy to recognize something we've seen before? MIT research may provide an answer. 40% of all web queries concern re-finding already known objects. Describing these objects with a query may be difficult, but recognizing them in a search result should be easy. And we can make it even easier by presenting image thumbnails and other salient features alongside the usual titles and teasers.

Dan Russell from Google’s search quality team has a post at the Google Blog on the art of field study. In short, he describes how the search interface plays a role in search quality, and what measures Google takes to study and improve the user interface, to ultimately help improve search quality.