CityU USTC Advanced Research Institute

Suzhou, China

CityU USTC Advanced Research Institute

Suzhou, China
SEARCH FILTERS
Time filter
Source Type

Ni X.,Shanghai University | Ni X.,Hefei University of Technology | Ni X.,City University of Hong Kong | Ni X.,CityU USTC Advanced Research Institute | And 10 more authors.
Information Processing and Management | Year: 2012

In this paper, we propose a generative model, the Topic-based User Interest (TUI) model, to capture the user interest in the User-Interactive Question Answering (UIQA) systems. Specifically, our method aims to model the user interest in the UIQA systems with latent topic method, and extract interests for users by mining the questions they asked, the categories they participated in and relevant answer providers. We apply the TUI model to the application of question recommendation, which automatically recommends to certain user appropriate questions he might be interested in. Data collection from Yahoo! Answers is used to evaluate the performance of the proposed model in question recommendation, and the experimental results show the effectiveness of our proposed model. © 2011 Elsevier Ltd. All rights reserved.


Wen S.,Hefei University of Technology | Wen S.,CityU USTC Advanced Research Institute | Wen S.,City University of Hong Kong | Li Q.,CityU USTC Advanced Research Institute | And 11 more authors.
Service Oriented Computing and Applications | Year: 2012

For a number of services with similar functionality reputation has been regarded as one of the most important methods to identify good ones from bad ones. However, a composite service, which is composed of multiple component services, obtains only one score (or feedback) after every invocation. In order to compute the reputation of each component service, it is necessary for the composite service to distribute this score to its component services. How to achieve a fair distribution is a challenging issue, as each component service may perform differently in contributing to the success or failure of the composite service. Although several efforts have been made for this problem, they do not consider the context of composition, which makes the distribution unfair. Therefore, in this paper, we propose a fair score distribution framework which combines the context of component services and their runtime performance. We distinguish two aspects contexts of a component service: structure-related importance and community-related replaceability, and adopt graph theory and dominating relationship technique to compute them, respectively. Experimental results show that our approach can achieve a more reasonable and fair score distribution than other existing methods. © 2012 Springer-Verlag London Limited.


Song W.,Hefei University of Technology | Song W.,City University of Hong Kong | Song W.,CityU USTC Advanced Research Institute | Wenyin L.,City University of Hong Kong | And 5 more authors.
Information Processing and Management | Year: 2011

Question categorization, which suggests one of a set of predefined categories to a user's question according to the question's topic or content, is a useful technique in user-interactive question answering systems. In this paper, we propose an automatic method for question categorization in a user-interactive question answering system. This method includes four steps: feature space construction, topic-wise words identification and weighting, semantic mapping, and similarity calculation. We firstly construct the feature space based on all accumulated questions and calculate the feature vector of each predefined category which contains certain accumulated questions. When a new question is posted, the semantic pattern of the question is used to identify and weigh the important words of the question. After that, the question is semantically mapped into the constructed feature space to enrich its representation. Finally, the similarity between the question and each category is calculated based on their feature vectors. The category with the highest similarity is assigned to the question. The experimental results show that our proposed method achieves good categorization precision and outperforms the traditional categorization methods on the selected test questions. © 2010 Elsevier Ltd. All rights reserved.


Quan X.,City University of Hong Kong | Liu G.,City University of Hong Kong | Lu Z.,City University of Hong Kong | Ni X.,City University of Hong Kong | And 4 more authors.
Knowledge and Information Systems | Year: 2010

In this paper, we propose a new method for measuring the similarity between two short text snippets by comparing each of them with the probabilistic topics. Specifically, our method starts by firstly finding the distinguishing terms between the two short text snippets and comparing them with a series of probabilistic topics, extracted by Gibbs sampling algorithm. The relationship between the distinguishing terms of the short text snippets can be discovered by examining their probabilities under each topic. The similarity between two short text snippets is calculated based on their common terms and the relationship of their distinguishing terms. Extensive experiments on paraphrasing and question categorization show that the proposed method can calculate the similarity of short text snippets more accurately than other methods including the pure TF-IDF measure. © 2009 Springer-Verlag London Limited.


Shi L.,CityU USTC Advanced Research Institute | Shi L.,Hefei University of Technology | Shi L.,City University of Hong Kong | Xue C.J.,City University of Hong Kong | Zhou X.,Hefei University of Technology
Real-Time Technology and Applications - Proceedings | Year: 2011

Flash memory is becoming the storage media of choice for mobile devices and embedded systems. The performance of flash memory is impacted by the asymmetric speed of read and write operations, limited number of erase times and the absence of in-place updates. To improve the performance of flash memory based systems, write buffer cache has been equipped in flash memories recently. At the same time, new virtual memory management strategies have been proposed in recent studies that consider the characteristics of flash memory. In this paper, we propose cooperative management schemes for virtual memory and write buffer cache to improve performance of flash memory based systems. Management on virtual memory is designed to exploit the write buffer cache status and management on write buffer cache is designed to cooperate with virtual memory to improve flash memory performance. The experimental results show that the proposed approaches can improve I/O performance by 34.8% on average, and reduce the number of erase and write operations by 44.7% and 28.6% on average. © 2011 IEEE.


Ni X.,Hefei University of Technology | Ni X.,City University of Hong Kong | Ni X.,CityU USTC Advanced Research Institute | Quan X.,City University of Hong Kong | And 6 more authors.
Knowledge and Information Systems | Year: 2011

A new clustering strategy, TermCut, is presented to cluster short text snippets by finding core terms in the corpus. We model the collection of short text snippets as a graph in which each vertex represents a piece of short text snippet and each weighted edge between two vertices measures the relationship between the two vertices. TermCut is then applied to recursively select a core term and bisect the graph such that the short text snippets in one part of the graph contain the term, whereas those snippets in the other part do not. We apply the proposed method on different types of short text snippets, including questions and search results. Experimental results show that the proposed method outperforms state-of-the-art clustering algorithms for clustering short text snippets. © 2010 Springer-Verlag London Limited.


Ye F.,Hefei University of Technology | Ye F.,CityU USTC Advanced Research Institute | Ye F.,City University of Hong Kong | Li Q.,CityU USTC Advanced Research Institute | And 3 more authors.
World Wide Web | Year: 2011

Mobile Peer to Peer (MP2P) networks provide decentralization, self-organization, scalability characters, but suffer from high latency and link break problems. In this paper, we study the cache/replication placement and cache update problems arising in such kind of networks. While researchers have proposed various replication placement algorithms to place data across the network to address the problem, it was proven as NP-hard. As a result, many heuristic algorithms have been brought forward for solving the problem. In this article, we propose an effective and low cost cache placement strategy combined with an update scheme which can be easily implemented in a decentralized way. The contribution of this paper is the adaptive and flexible cache placement and update algorithms designed for real MP2P network usage. The combination of MP2P cache placement and update is the novelty of this article. Extensive experiments are conducted to demonstrate the efficiency of the cache placement and update scheme. © 2010 Springer Science+Business Media, LLC.


Lu Y.,Hefei University of Technology | Lu Y.,City University of Hong Kong | Lu Y.,CityU USTC Advanced Research Institute | Kong X.,City University of Hong Kong | And 5 more authors.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2010

Existing research efforts in sentiment analysis of online user reviews mainly focus on extracting features (such as quality and price) of products/services and classifying users' sentiments into semantic orientations (such as positive, negative or neutral). However, few of them take the strength of user sentiments into consideration, which is particularly important in measuring the overall quality of products/services. Intuitively, different reviews for the same feature should have quite different sentiment strength, even though they may express the same polarity of sentiment. This paper presents an approach to estimating the sentiment strength of user reviews according to the strength of adverbs and adjectives expressed by users in their opinion phrases. Experimental result on a hotel review dataset in Chinese shows that the proposed approach is effective in the task of sentiment classification and achieves a good performance on a multi-scale evaluation. © 2010 Springer-Verlag.


Liu A.,Hefei University of Technology | Liu A.,City University of Hong Kong | Liu A.,CityU USTC Advanced Research Institute | Li Q.,City University of Hong Kong | And 3 more authors.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2011

Web services cannot be always available as they are typically deployed in a dynamic environment. As an effective approach to improving Web services availability, replication has received much attention recently. How to design an optimal replication scheme with the best QoS, however, remains an open problem due to its inherent computational hardness. In this paper, we propose an efficient approach to designing a near-optimal replication scheme. We adopt directed acyclic graph (DAG) as the modeling tool for replication scheme and then utilize DAG coding for performance optimization. Simulation results show our approach can generate a near-optimal replication scheme with acceptable computation overheads. © 2011 Springer-Verlag.


Liu A.,Hefei University of Technology | Liu A.,CityU USTC Advanced Research Institute | Liu A.,Wuhan University | Li Q.,City University of Hong Kong | And 3 more authors.
Proceedings - 2012 IEEE 19th International Conference on Web Services, ICWS 2012 | Year: 2012

Reputation is useful for establishing trust between Web service (WS) providers and WS consumers. In the context of WS composition, a challenging issue of reputation management is to propagate a user's impression of a composite WS (i.e., the user's feedback rating) to its component WSs. In this paper, we propose a Shapley value based approach which can achieve fair impression propagation, that is, the reputation of a component WS is never awarded (or penalized) for the good (or bad) performances of the other peer component WSs in the same composite WS. The fairness of the proposed approach is validated through theoretical analysis and experimental results. © 2012 IEEE.

Loading CityU USTC Advanced Research Institute collaborators
Loading CityU USTC Advanced Research Institute collaborators