This is the second time we participate in TREC Blog Track. There are three main tasks in the track, relevant finding task, opinion finding task and polarity task. In this year, we use multi-field relevance ranking in ...
详细信息
This is the second time we participate in TREC Blog Track. There are three main tasks in the track, relevant finding task, opinion finding task and polarity task. In this year, we use multi-field relevance ranking in relevant finding task;and in opinion finding task, we focused on the combination of relevance score and opinionate score use a unified generation model;in polarity task, we develop two new methods to find out positive and negative blogs.
Our group has participated into Relevance Feedback (RF) Track in TREC2008. In our experiments, two kinds of techniques, query expansion and search result re-ranking based on document relevance model, are adopted to im...
详细信息
Our group has participated into Relevance Feedback (RF) Track in TREC2008. In our experiments, two kinds of techniques, query expansion and search result re-ranking based on document relevance model, are adopted to improve the retrieval performance. The TMiner search engine, from IR group of Tsinghua University, is used as our text retrieval system.
Hot events reveal the society's attention and can help to watch the public voice or make decisions. Currently, hot events of various domains are provided by some organizations. Most of the results are considered t...
详细信息
Hot events reveal the society's attention and can help to watch the public voice or make decisions. Currently, hot events of various domains are provided by some organizations. Most of the results are considered to be generated manually, which are subjective though costing lots of human efforts. In this paper, a two-stage automatic algorithm for hot event detection is proposed. The first stage is event detection and tracking, in which news stories are organized into events accurately and effectively. The second stage is event ranking, in which both the media and user attention paid to events are taken into consideration. Top ranked events are returned as hot ones. Experiments performed on practical Web datasets show that: (1) F-measure of the best event detection and tracking results is close to 90%;(2) Hot events generated by the proposed algorithm reflect both the media and users' influence on the event ranking, which is considered to be more objective.
A normalized measure is established to provide the quantitative information about the degree of observability for the discrete-time, stochastically autonomous system. This measure is based on the generalized informati...
详细信息
A normalized measure is established to provide the quantitative information about the degree of observability for the discrete-time, stochastically autonomous system. This measure is based on the generalized information theoretic quantities （generalized entropy, mutual information） of the system state and the observations, where the system state can be a discrete or a continuous random vector. Some important properties are presented. For the linear case, the explicit formula for the degree of observability is derived, and the equivalence between the proposed measure and the traditional rank condition is proved. The curves for the degree of observability are depicted in a simple example.
In this paper, we present a novel method for human action recognition with the combined global movement feature and local configuration feature. The human action is represented as a sequence of joints in the 4D spatio...
Rescue robot places an important role when any severe disaster in underground coal mine happens. To locate and keep a track of miners working in a well becomes an impossible completed mission once the well communicati...
详细信息
We participate in document search and expert search of Enterprise Track in TREC2008. The corpus and tasks are same as the year before. Different from TREC 2007, the topics come from CSIRO Enquiries, and the topic stat...
详细信息
We participate in document search and expert search of Enterprise Track in TREC2008. The corpus and tasks are same as the year before. Different from TREC 2007, the topics come from CSIRO Enquiries, and the topic statements are richer and more colloquial.. In document search, we look into the key resource page pre-selection, the use of anchor text, query classification, and multi-field search. In expert search, we develop methods to detect expert identifiers and experimented based on our previous PDD (personal description documents) model.
Classifying what-type questions into proper semantic categories is found more challenging than classifying other types in question answering systems. In this paper, we propose to classify what-type questions by head n...
详细信息
We introduce a new multi-player geometric game, which we will refer to as the isolation game, and study its Nash equilibria and best or better response dynamics. The isolation game is inspired by the Voronoi game, com...
详细信息
In this paper, we propose a novel approach to organize image search results obtained from state-of-the-art image search engines in order to improve user experience. We aim to discover exemplars from search results and...
详细信息
ISBN:
(纸本)9781605583037
In this paper, we propose a novel approach to organize image search results obtained from state-of-the-art image search engines in order to improve user experience. We aim to discover exemplars from search results and simultaneously group the images. The exemplars are delivered to the user as a summary of search results instead of the large amount of unorganized images. This gives the user a brief overview of search results with a small amount of images, and helps the user to further find the images of interest. We adopt the idea of affnity propagation and design a fast sparse affnity propagation algorithm to find exemplars that best represent the image search results. Experiments on real-world data demonstrate the effectiveness of our method both visually and quantitatively. Copyright 2008 ACM.
暂无评论