Object-level Vertical Search

Current Web search engines can be considered page-level general search systems with the main function of ranking web pages according to the relevance of given queries. However, we see an opportunity of further improving search technologies by developing specialty search engines that go deeper into vertical domains to provide more precise answers in search results. Vertical search will also be able to provide more relevant and intelligent search results since the method can data-mine associations and relations within data. In this talk, I will introduce the current status of web search and discuss this new trend toward building what I call "object-level vertical search engines." The research problems we need to address include large- scale web classification, structured data extraction, and information aggregation and integration from a variety of sources, including data feeds and crawl. I will introduce recent research progress at MSR Asia and show how we are applying our developed technologies to build more advanced vertical search engines for such areas as shopping, academic search, and multimedia.