[Edit]

Gengxin Miao [FOAF]  [Follow]

Position: Brief Introduction: In China, the main officials in local government get re-voted every five years. Traditionally, the delegates of citizens fill cards to vote all the positions. Then, the cards are read manually. This is a really heavy and boring task for human. This system can read and calculated the information automatically by means of image processing. Computers are fixed in the voting bins. Mechanical and electrical devices make sure that all the vote cards stop exactly in the same place and then an image of the vote card is captured and sent to the computer. A program processed the image and calculate the gain of each candidate respectively. Since the voting process is carried out in a very large hall and there are a large number of citizen delegates, there are several vote bins working in parallel. There is a star-like bus connected all the bins; a center computer combine all the vote data together and give the final vote results
Affiliation: * 2004.9~now Research Assistant in Dept. of Automation, THU
Address: Xing Xie, Ruihua Song Brief Introduction: Currently, most of the search engines use the same retrieval model when processing all the queries submitted by users. However, when we evaluate the performance of the search engine based on every single query, we found that they are quite different from each other. Based on this observation, we found that different queries should be processed with different retrieval models. Some former research results show that queries can be classified based on the users' different information needs. But is this the optimal query classification strategy? Should queries with different information needs with different retrieval techs? This project aims at solve these problems. We try to divide the queries into several categories based on the optimal retrieval tech which should be used. And then several sub-models are trained for each category of queries. When new query comes, we assign the adaptive model to do the retrieval. * 2004.9~now Night View People Involved: Gengxin Miao, Prof. Yupin Luo, LYP Group
Homepage: http://learn.tsinghua.edu.cn/homepage/20...
[Edit]
Statistics: H-index: 3 (See all experts' h-index.)
total citation number: 38
highest-cited paper: Efficient Browsing of Web Search Results on Mobile Devices Based on Block Importance Model (2005) at PerCom (Cited By 28)

Research Interest:

Collaborative Web Data Record, Pedestrian Detection System, Web Search Results, e-Healthcare System, Block Importance Model

Show Temporal Interests (Do you want to see the change of his/her research interests?)


Publications: [Edit disambiguation Result]