|
iDEA: Drexel E-repository and Archives >
Drexel Academic Community >
College of Information Science and Technology >
Research Day Posters (IST) >
HMM-based artificial designer for search interface segmentation
Please use this identifier to cite or link to this item:
http://hdl.handle.net/1860/3713
|
| Title: | HMM-based artificial designer for search interface segmentation |
| Authors: | Khare, Ritu
An, Yuan
Song, Il-Yeol |
| Keywords: | User interfaces (Computer systems)
Web search engines
Hidden Markov Models |
| Issue Date: | 23-Apr-2009 |
| Series/Report no.: | IST Research Day 2009 posters |
| Abstract: | Search interfaces are the primary "doors" to the data-rich and continually growing deep Web. Hence, their understanding is a prerequisite to the design of applications that are dedicated to make deep Web contents more useful. The problem of automatic search interface understanding has recently been addressed by several works using techniques such as rules, heuristics, and even machine learning. However, a very challenging portion (segmentation) of this problem has not received appropriate attention. In this poster, we present an approach to segment a search interface into logical attributes and assign semantic labels to the interface components. The key feature in our approach is a 2-layered hidden Markov model (HMM), which encodes the learned knowledge about designing a search interface for querying underlying databases. We tested our approach on the less-explored family of search interfaces belonging to scientific databases and found promising results. On comparison with an existing approach, our approach improved the segmentation accuracy by approximately 10%. |
| URI: | http://hdl.handle.net/1860/3713 |
| Appears in Collections: | Research Day Posters (IST)
|
Items in iDEA are protected by copyright, with all rights reserved, unless otherwise indicated.
|
www.drexel.edu
