Title: | Semantic-based Language Models for Information Retrieval and Text Mining |
Speaker: | Prof. Tony Hu, Drexel University, USA |
Time/Place: | 10:30 - 12:30 FSC1217, Fong Shu Chuen Library, HSH Campus, Hong Kong Baptist University |
Abstract: | The language modeling approach centers on the issue of estimating an accurate model by choosing appropriate language models as well as smoothing techniques. In the thesis , we propose a novel context-sensitive semantic smoothing method referred to as a topic signature language model. It extracts explicit topic signatures from a document and then statistically maps them into individual words in the vocabulary . In order to support the new language model, we developed two automated algorithms to extract multiword phrases and ontological concepts, respectively, and an EM-based algorithm to learn semantic mapping knowledge from co-occurrence data. The topic signature language model is applied to three applications: information retrieval, text classification, and text clustering. The evaluations on news collection and biomedical literature prove the effectiveness of the topic signature language model. In the experiment of information retrieval, the topic signature language model consistently outperforms the baseline two-stage language model as well as the context-insensitive semantic smoothing method in all configurations. It also beats the state-of-the-art Okapi models in all configurations. In the experiment of text classification, w hen the size of training documents is small, the Bayesian classifier with semantic smoothing not only outperforms the classifiers with background smoothing and Laplace smoothing , but it also beats the active learning classifiers and SVM classifiers. On the task of clustering, whether or not the dataset to cluster is small, the model-based k-means with semantic smoothing performs significantly better than both the model-based k-means with background smoothing and Laplace smoothing . It is also superior to the spherical k-means in terms of effectiveness. In addition, we empirically prove that, within the framework of topic signature language models, the semantic knowledge learned from one collection could be effectively applied to other collections. In the thesis, we also compare three types of topic signatures (i.e., words, multiword phrases, and ontological concepts), with respect to their effectiveness and efficiency for semantic smoothing. In general, it is more expensive to extract multiword phrases and ontological concepts than individual words, but semantic mapping based on multiword phrases and ontological concepts are more effective in handling data sparsity than on individual words. |
Title: | ICM Distinguished Computational Mathematics Lecture: On the Computation of Highly Oscillatory Equations |
Speaker: | Prof. Arieh Iserles, Department of Applied Mathematics and Theoretical Physics, University of Cambridge, UK |
Time/Place: | 10:45 - 11:45 (Preceded by Reception at 10:30am) NAB209, Lam Woo International Conference Center, Shaw Campus, HKBU |
Abstract: | In this talk we consider efficient computation of differential and differential-algebraic systems that originate in the modelling of electronic circuits. We represent the solution as an asymptotic series where each consecutive term is a modulated Fourier expansion. This reduces the computation to that of nonoscillatory systems, in tandem with the evaluation of modulated Fourier expansions with FFT. This methodology has the welcome feature that, while the cost of the computation is independent of frequency, its accuracy improves the more rapidly the solution oscillates. |
Title: | CMIV and ICM Lecture: Web-based Question-Answering |
Speaker: | Prof. Tony Hu, Drexel University, USA |
Time/Place: | 10:30 - 12:30 DLB614, David C. Lam Building, Shaw Campus, Hong Kong Baptist University |
Abstract: | In this presentation, I briefly review the state-of-the-art approaches to web-based factoid question answering and then introduce the new approaches used in a new QA system called DragonQA. The DragonQA system has a lightweight implementation; it does not require intensive linguistic resources. It proposes a novel efficient method to score the relevance of candidate answers to the original question based on the local context. Basically, it evaluates the similarity of candidate sentences to the question; if the similarity is greater than the set threshold, the distance of extracted candidates to the query words in the sentence will be further evaluated. The new scoring method does not require any domain-specific knowledge and thus can be applied to open domain questions. A formal evaluation on the TREC QA dataset showed that the DragonQA outperformed significantly two state-of-the-art QA systems, ARANEA and QUANTA. |
Title: | SRCC DLS: What is a good risk measure? |
Speaker: | Prof. Jia-An Yan, Chinese Academy of Sciences, China |
Time/Place: | 11:00 - 12:30 (Preceded by Reception at 10:30am) ACC109, Jockey Club Academic Community Centre, Baptist University Road Campus, HKBU |
Abstract: | In this talk, I first show why VaR (value at risk), a popular risk measure, is not a good risk measure. Then I present various new static risk measures: risk measures with comonotonic subadditivity or convexity, law-invariant risk measures, risk measures respecting stochastic orders. Finally, we show that some set of axioms for risk measures are equivalent, and conclude that a law-invariant coherent (or convex) risk measure is a good risk measure. |
Title: | PL: U.S. Supreme Court: Facts and Figures |
Speaker: | Prof. Franklin Luk, Hong Kong Baptist University, Hong Kong |
Time/Place: | 16:15 - 17:15 (Preceded by Reception at 15:45pm) WLB203, The Wing Lung Bank Building, Shaw Campus, HKBU |
Abstract: | There are three parts in this presentation: 1. A brief introduction to the U.S. Supreme Court is given, with particular emphasis on its composition. A surprising fact is that the number of Supreme Court Justices has not always been nine; indeed, it was never fixed as an odd number. At various times, there were six, eight, and ten Justices. 2. A 5-4 Supreme Court vote in 2000 that selected George W. Bush as the next U.S. President is described in some detail. It was a vote that would impact world politics and finances. 3. The Second Rehnquist Court (1994-2005) is chosen for a case study. This Court had the second longest reign in U.S. history, and the longest in over 180 years. The following Mathematical Problem is discussed: "Construct a low-rank approximation of the voting pattern that preserves the voting margins." |
Title: | The continuous and discontinuous Galerkin finite element methods for linear delay differential equations |
Speaker: | Prof. Chengjian Zhang and Prof. Dongfang Li, School of Mathematics and Statistics, , Huazhong University of Science and Technology, China |
Time/Place: | 11:30 - 12:30 FSC1217, Fong Shu Chuen Library, HSH Campus, Hong Kong Baptist University |
Abstract: | This talk deals with the continuous and discontinuous Galerkin finite element methods for linear delay differential equations. By the element-orthogonal analytical method (see e.g. [1]), the superconvergence results of the both methods are derived at nodal points and eigenpoints, respectively. Numerical experiments confirm the methods' effectiveness and the theoretical results. References [1] C. Chen, Structure Theory of superconvergence of Finite Elements, Hunan Press of Science and Technology, Changsha, 2001. To modify the abstract, paste the updated one to here again! |
We organize conferences and workshops every year. Hope we can see you in future.
Learn MoreProf. M. Cheng, Dr. Y. S. Hon, Dr. K. F. Lam, Prof. L. Ling, Dr. T. Tong and Prof. L. Zhu have been awarded research grants by Hong Kong Research Grant Council (RGC) — congratulations!
Learn MoreFollow HKBU Math