Bernard Mangold writes, about yahoo's newest offering : A new twist on search that uses machine learning technology Yahoo!mindset. Search results sorted according to whether they are more commercial or more informational (i.e., from academic, non-commercial, or research-oriented sources). Sometimes you want to buy stuff and sometimes you just want to do research. In a typical search page, results point to commercial pages that are mixed together with non-commercial pages, so it's harder to find the type of information you're looking for. Mindsethelps solve that problem.
Through the use of machine learning for text classification, yahoo classifies each web page in the top 100 search results for a query. Then sorted those results according to the preference set. Mindset uses an intuitive slider in the interface, so the user can set the bias for commercial vs. non-commercial results. Pages not classified are tried to be classified at the background. It allows user to sort search results for your query into commercial or non-commercial (informational) results, based on whether on the need : shopping or seeking information. Every page is assigned a relatively continuous score ranging from -2 (most commercial) to +2 (most informational). Pages scored 0 are a balance of commercial and informational.
Glinden says he is surpised to see this focus on sliders. He thinks that these aren't particularly useful. Most novice users will not use or understand the slider; they just want the top result to be useful. It's not even that useful to power users since sliders fail to provide the level of granularity they need. More knobs, more buttons, more complexity. With less than 1% of users even bothering with the existing advanced search options on search engines, are more controls really what people need? Most searchers just want the right thing to happen. They want it to just work. Me think - this is a good advancement - for one it brings lot ore transparency in ranking the pages and fundamentally recognises that different clases of users or at different times the expectations on search results could be different - good attempt in that without system guessing what the user may be needing based on profile inforation which may not be contextual - this looks better.
|
