Japanese is here.
Abstract Search engine result pages, aka SERP, provide page titles, URLs, and text based snippets of the searched pages. These help users to decide to click or not each searched page. Text based snippets, however, are different from the real page rendered by web browsers, after an user clicks one of the searched pages, the user may see an unexpected one. Even if the page is the expected one, the user cannot find immediately where the information they want to know is described in the page. In this paper, we propose visual patch, which is provided as a graphical snippet in SERP and is rendered by web browsers already. Visual patch is a visual context based snippet, not a thumbnail of the whole web page in a smaller size, and is trimmed around the occurrence of the query keyword in the same size as the original page rendered by web browsers. Users can see the layouts around the keyword and read the concrete text just before clicking the searched pages. Thus visual patch in SERP shows not web pages which may include the information users want to know, but the information they want to know or not.