An Aggregate Search Model for Web Search Engines: An Empirical Study

In this paper, we study a novel aggregate search model for web search engines. Rather than retrieving individual web pages in the search result, our model aggregates relevant web pages and formulates information groups which may capture user's search intents well. An information group may consist of an individual web page, or a set of hyper-linked web pages that are relevant to user's queries. Several meaningful ranking measures are proposed to rank returned information groups. We evaluate the proposed aggregate search model using a large real search log dataset and an open source web search platform. The empirical study indicates that our model is useful to improve the web search quality.