| 摘要 |
A search term is generated based on an object containing a plurality of items. The object may be a text object 110, e.g. a document, containing text items 111. At least one, but not all, of the text items are selected based on a user-defined profile 108 and parsed to generate the search term. Alternatively, the object may be a non-text object 112, 114, 116, such as a URL or a JPEG, TIFF or WAV file, containing an item that is identifiable, such as a person's face, a car, a sound, a word, etc. The item is parsed to generate a text representation of the item based on the profile and search terms are generated based on the text representation. The parser 104 may parse selected items by eliminating common words, ranking words, or based on terms in the profile. Alternatively, the profile may contain an instruction to identify, for example, a car in an image object. Search terms may be sent to a search engine 106 to generate search results. A user may drag and drop a graphical representation of the object into a search engine icon to automatically perform a search based on the profile. |
