資料探勘與知識開發.pptVIP

* * * * * * * * * * * * * * Usage mining Application events Atomic application event to complex application event Atomic application event Complex application event Usage mining Application events Atomic application event to complex application event People who bought A also bought B Loop ? Use mining to get semantic Use semantic to help mining Closing the loop Purpose To generate metadata that reflect a semantic model underlying the site To identify both in the pages text and in their usage and bases on these insights to improve information architecture and page design Step 1 ontology learning Step 2 the ontology field Step 3 knowledge is mined Step 4 domain level usage profiles Domain level usage profiles Get the reason behind the pattern Three hotels in the same category but we don’t known why Result New category : golf hotel Site model has been achieved Ex: search accommodation in zingst Hotel is subclass of accommodation Zingst and Ahrenshoop on the same peninsula Enable search engines and other programs to better understand the content of Web pages and sites Web mining methods are able to understand and shape the Web Conclusion and Outlook * * * * * * * * * * * * * * * * * Semantic Web Mining State of the Art and Future Directions Gerd Stumme, Andreas Hotho, Bettina Berendt Outline Introduction Semantic Web Web Mining Extracting Semantic from the Web Using Semantics for Web Mining and Mining the Semantic Web Closing the Loop Conclusion and Outlook The amount of data is so huge Semantic Web addresses the first part Web Mining addresses the second part Aims at combining the two areas Semantic Web and Web Mining Introduction Question 1 : Machine support is limited Question 2 : Now too often excessively large or inadequate lists of hits Question 3 : Large data is interpretable Semantic Web (1/9) What are the semantic web direction？ Providing a common syntax for machine understandable statements Establishing common vocabularies. Agreeing on a logical l