Presenting compelling search results depends critically on understanding what is there to be presented on the first place. Given that the current generation of search engines have a very limited understanding of the query entered by the user, the content returned as a result and the relationship of the two, the opportunities for customizing search results have been limited.

Opening up search
The majority of Web pages today are generated from databases, and Web site owners increasingly are providing APIs to this data or embedding information inside their HTML pages with microformats, eRDF, or RDFa. In other cases, structured data can be extracted with relative ease from Web pages that follow atemplate using XSLT stylesheets.

SearchMonkey reuses structured data to improve search result display with benefits to both search users, developers, and publishers of web content. The first type of applications are focusing on remaking the abstracts on the search result page: Figure 1 shows the kind of presentations that structured data enables in this space. Based on data, the image representing the object can be easily singled out. One can also easily select the most important attributes of the object to be shown in a table format. Similarly for links: the data tells which links represent important actions the user can take (e.g. play the video, buy the product) and these links can be arranged in a way that their function is clear. In essence, knowledge of the data and its semantics enables to present the page in a much more informative, attractive, and concise way.

The benefits for publishers are immediately clear: when presenting their page this way publishers can expect more clicks and higher quality traffic flowing to their site. In fact, several large publishers have moved to implement semantic metadata markup (microformats, eRDF, RDFa) specifically for providing data to SearchMonkey.

On the other hand, users also stand to benefit from a better experience. The only concerns on the users’ part is the possibility of opting out and having a system free of spam. Both concerns are addressed by the Yahoo Application Gallery. As shown in Figure 2, the Gallery allows users to selectively opt in to particular SearchMonkey applications. Users also have a small number of high-quality applications that are turned on by default.

The gallery is also an effective spam detection mechanism: applications that are popular are unlikely to contain spam. Also important to note that the presence of metadata does not affect the ranking of pages. Pages that are trusted by the search engine based on other metrics can be also expected to contain trustable metadata.

Download pdf Semantic Web Use Cases and Case Studies