400 likes | 409 Views
Explore the concept of desktop search, including its importance, limitations, and ongoing research. Discover how context and user behavior can improve search accuracy and efficiency. Gain insight into the challenges and potential solutions in desktop search.
E N D
An Introduction to Desktop Search 寇玉波 Web组
Outline • A Story • What is Desktop Search • Background • Research on Desktop Search • Conclusion
Outline • A Story • What is Desktop Search • Background • Research on Desktop Search • Conclusion
A story that happened in 2007: Google complains to DOJ about Vista search ↓ DOJ Pushes Microsoft to Produce Vista SP1 in '07 ↓ Microsoft released Service Pack 1 for Vista ↓ Google says Vista search changes not enough DOJ : Department of Justice
Outline • A Story • What is Desktop Search • Background • Research on Desktop Search • Conclusion
Desktop search has become important for several reasons • Analysts believe Web surfers are pining for a "universal search" solution • a user who becomes loyal to a desktop search product is highly likely to extend that loyalty to the tool maker's Internet search engine. • increasingly users expect to be able to find information on their PCs in the same way they find information on the Internet.
Outline • A Story • What is Desktop Search • Background • Research on Desktop Search • Conclusion
Desktop search tools 2004 and 2005
Limitations: • fall short of utilizing desktop specific information, especially context information. • sophisticated ranking using PageRank and other features is unavailable on our desktop.
What is Context? Context is any information that can be used to characterize the situation of an entity. An entity is a person, place, or object that is considered relevant to the interaction between a user and an application, including the user and applications themselves.
Outline • A Story • What is Desktop Search • Background • Research on Desktop Search • Conclusion
Main Researchers: Paul - Alexandru Chirita Wolfgang Nejdl 1, Integrating Context Metadata within Desktop Search 2, Analyzing User Behavior to Rank Desktop Items 3, Pushing Task Relevant Web Links down to the Desktop
Main Researchers: Paul - Alexandru Chirita Wolfgang Nejdl 1, Integrating Context Metadata within Desktop Search 2, Analyzing User Behavior to Rank Desktop Items 3, Pushing Task Relevant Web Links down to the Desktop
I want to find the attachment The cat sent me in his reply-email reply attachment Retrieve the correct document by using body text and sender information associated to this document, inherited from the original email
additional information provided through the metadata associated with the chosen result document
Main Researchers: Paul - Alexandru Chirita Wolfgang Nejdl 1, Integrating Context Metadata within Desktop Search 2, Analyzing User Behavior to Rank Desktop Items 3, Pushing Task Relevant Web Links down to the Desktop
Web search has become more efficient than PC search due to the powerful link based ranking solutions like PageRank
in almost all cases when two items are touched in a sequence several times, there will also be a relation between them, irrespective of the underlying user activity. b a c a Link structure ………… a b b b a
Other Heuristics to Generate Desktop Links • the files stored within the same directory • files having the same file name (ignoring the path) • the resources sharing the same type Link structure
Main Researchers: Paul Alexandru Chirita Wolfgang Nejdl 1, Integrating Context Metadata within Desktop Search 2, Analyzing User Behavior to Rank Desktop Items 3, Pushing Task Relevant Web Links down to the Desktop
automatically search the web and recommend URLs relevant to user’s current work query result software agent extract recommend URLs…….. URLs……. URLs…….
Exploiting only the Currently Active DocumentvsExploiting the Full Context of the Currently Active Document ………
EXTRACTING RELEVANT QUERY KEYWORDS Term Frequency (TF) Sentence Selection (SS) Lexical Compounds (LC)
Term Frequency (TF) TF TermScore 1 Word1 10 10 2 Word2 6 5.2 3 Word3 6 4.5 4 Word4 4 2
EXTRACTING RELEVANT QUERY KEYWORDS Term Frequency (TF) Sentence Selection (SS) Lexical Compounds (LC)
Main Researchers: Paul - Alexandru Chirita Wolfgang Nejdl 1, Integrating Context Metadata within Desktop Search 2, Analyzing User Behavior to Rank Desktop Items 3, Pushing Task Relevant Web Links down to the Desktop
Outline • A Story • What is Desktop Search • Background • Research on Desktop Search • Conclusion
Conclusion and Next Step • Ideas about rank and utilizing contextual information to cope with current limitations • Compare current desktop search tools and find more knowledge • How to better support search in personal dataspace
References: • Activity Based Metadata for Semantic Desktop Search • Analyzing User Behavior to Rank Desktop Items • Beagle++: Semantically Enhanced Searching and Ranking on the Desktop • Building a Desktop Search Test-bed • Desktop Context Detection Using Implicit Feedback • Desktop Search-How Contextual Information Influences Search Results & Rankings • I know I stored it somewhere -Contextual Information and Ranking on our Desktop • Pushing Task Relevant Web Links down to the Desktop • The beagle++ toolbox_ Towards an extendable desktop search architecture • Introduction to Cloud Computing • http://en.wikipedia.org/wiki/Desktop_search • http://www.google.com/search?hl=en&q=desktop+search&btnG=Google+Search • http://www.baidu.com/s?wd=%D7%C0%C3%E6%CB%D1%CB%F7