08-08-2012, 02:59 PM
Information Retrieval in Mobile Phones using Snippet Clustering Methods
[attachment=30603]
Abstract
Mobile web refers to access to the web from a mobile computing system.
The access to the web from a mobile device for the search of information is growing in popularity due to the availability, cheap cost and convenience of access.
Due to the potentially large volume of content that can be accessed, low power limitations of cellular band width, variable and small screen sizes pose considerable challenges to mobile users.
so this paper proposes a middleware based system that retrieves content dynamically from the web based on the user context, personal preferences and uses adapted snippet clustering algorithms.
Introduction
The initial work has been made by the search engine developers through mobile based customized search applications.
This work is notable development in terms of stable user interfaces based on device profiles.
Mobile device users need proper visualization methods , mobile keypad based user interfaces , fast response time and personalized content.
The search application can be modeled as a Middleware system.
The middleware system can abstract the implementation details from the user and create an environment of distraction free computing.
The design of middleware is a non-trivial task and can handle the complexities of the mobile phone based environment and provide a unified interface to the users.
This work proposes a middleware layer to
(a) Personalized access
(b) Formulate keypad based interaction mechanisms and
© Interface with search engines using adapted snippet clustering
algorithms.
Related work
The related work can be summarized in terms of
a)device manufacturers and search engine approaches
b)middleware prototypes based on clients
c)server based middleware prototypes.
The work has proceeded in terms of query expansion, use of recommender systems and suggestions based on user interfaces.
There are many versions of search engines for mobile phones like yahoo mobile search , google mobile search.
These approaches do not offer any personaliztion , query expansion.
Adapted snippet clustering algorithm
1: Prune the snippet content by removing the stop words.
2. Use thesaurus to eliminate the snippet contents that are dynamically opposite to the given query.
3: Build a word- document index R={Siw1,Siw2,...,Siwk}, where Siwk contains the indexed documents of Snippet content Si.
4: Construct a similarity measure model and calculate the similarity matrix for user query, personal profile content and the related words with the Word document index R.
5: Form the suffix tree from the snippet content Sk.
Conclusion
The user’s navigation issues are easily handled by transforming the hyperlinks in the content in to option numbered items.
It helps in the smooth operation of the system.
The results show that the proposed approach with middleware interface built on top of existing search engines with user suggestion mechanisms and context based search is promising avenue for research.
The work in the future will consist of more detailed studies on the search accuracy, interface speed and the user navigation aspects.
In future the work will also be implemented in a server based system with the mobile middleware client merely forwarding the user input and displaying the results. This will enhance the overall system.
[attachment=30603]
Abstract
Mobile web refers to access to the web from a mobile computing system.
The access to the web from a mobile device for the search of information is growing in popularity due to the availability, cheap cost and convenience of access.
Due to the potentially large volume of content that can be accessed, low power limitations of cellular band width, variable and small screen sizes pose considerable challenges to mobile users.
so this paper proposes a middleware based system that retrieves content dynamically from the web based on the user context, personal preferences and uses adapted snippet clustering algorithms.
Introduction
The initial work has been made by the search engine developers through mobile based customized search applications.
This work is notable development in terms of stable user interfaces based on device profiles.
Mobile device users need proper visualization methods , mobile keypad based user interfaces , fast response time and personalized content.
The search application can be modeled as a Middleware system.
The middleware system can abstract the implementation details from the user and create an environment of distraction free computing.
The design of middleware is a non-trivial task and can handle the complexities of the mobile phone based environment and provide a unified interface to the users.
This work proposes a middleware layer to
(a) Personalized access
(b) Formulate keypad based interaction mechanisms and
© Interface with search engines using adapted snippet clustering
algorithms.
Related work
The related work can be summarized in terms of
a)device manufacturers and search engine approaches
b)middleware prototypes based on clients
c)server based middleware prototypes.
The work has proceeded in terms of query expansion, use of recommender systems and suggestions based on user interfaces.
There are many versions of search engines for mobile phones like yahoo mobile search , google mobile search.
These approaches do not offer any personaliztion , query expansion.
Adapted snippet clustering algorithm
1: Prune the snippet content by removing the stop words.
2. Use thesaurus to eliminate the snippet contents that are dynamically opposite to the given query.
3: Build a word- document index R={Siw1,Siw2,...,Siwk}, where Siwk contains the indexed documents of Snippet content Si.
4: Construct a similarity measure model and calculate the similarity matrix for user query, personal profile content and the related words with the Word document index R.
5: Form the suffix tree from the snippet content Sk.
Conclusion
The user’s navigation issues are easily handled by transforming the hyperlinks in the content in to option numbered items.
It helps in the smooth operation of the system.
The results show that the proposed approach with middleware interface built on top of existing search engines with user suggestion mechanisms and context based search is promising avenue for research.
The work in the future will consist of more detailed studies on the search accuracy, interface speed and the user navigation aspects.
In future the work will also be implemented in a server based system with the mobile middleware client merely forwarding the user input and displaying the results. This will enhance the overall system.