Vol. 33 · No. 3

Business processes are generally fixed and enforced strictly, as reflected by the static nature of underlying software systems and datasets. However, internal and external situations, organizational changes and various other factors trigger dynamism, which is reflected in the form of issues, complains, Q & A, opinions, reviews, etc., over a plethora of communication channels, such as email, chat, discussion forums, and internal social network. Careful and timely analysis and processing of such channels may lead to early detection of emerging trends, critical issues, opportunities, topics of interests, contributors, experts, etc. Social network analytics have been successfully applied in general purpose, online social network platforms, like Facebook and Twitter. However, in order for such techniques to be useful in business context, it is mandatory to integrate them with underlying business systems, processes and practices. Such integration problem is increasingly recognized as Big Data problem. We argue that Semantic Web technology applied with social network analytics can solve enterprise knowledge management, while achieving integration.

An acronym is an abbreviation of several words in such a way that the abbreviation itself forms a pronounceable word. Acronyms occur frequently throughout various documents, especially those of a technical nature, for example, research papers and patents. While these acronyms can enhance document readability, in a variety of fields, they have a negative effect on business intelligence. To resolve this problem, we propose a method of acronym-expansion disambiguation to collect high-quality enterprise information. In experimental evaluations, we demonstrate its efficiency through the use of objective comparisons.

This paper concerns the design and development of a distributed agent-based online system for English auctions. The proposed system is composed of two parts: an Agent-based Auction Server and a Web-based Graphical User Interface. The first part of our work brought about the advantages introduced by the multi-agent systems technology to the high-level of abstraction, modularity and performance of the server architecture and its implementation. On the server side, bids submitted by auction participants are handled by a hierarchical organization of agents that can be efficiently distributed on a computer network. This approach avoids the bottlenecks of bid processing that might occur during periods of heavy bidding, like for example snipping. We present experimental results that show a significant improvement of the server throughput compared with the architecture where a single auction manager agent is used for coordinating the participants for each active auction that is registered with the server. The second part of our work involved analysis of external functionalities, implementation and usability of a prototype online auction system that incorporates the Agent-based Auction Server. Our solution is outlined in terms of information flow management and its relation to the functionalities of the system. The main outcome of this part of the work is a clean specification of the information exchanges between the agent and non-agent software components of the system. Special attention is also given to the interoperability, understood here as successful integration of the different data communication protocols and software technologies that we employed for the implementation of the system.

In cross-enterprise collaborative environment, we have dealt with challenges in business process integration for common business goals. Research directions in this domain range from business to business integration (B2Bi) to service-oriented augmentation. Ontologies are used in Business Process Management (BPM) to reduce the gap between the business world and information technology (IT), especially in the context of cross enterprise collaboration. For a dynamic collaboration, virtual enterprises need to establish collaborative processes with appropriate matching levels of tasks. However, the problem of solving the semantics mismatching is still not tackled or even harder in case of querying space between different enterprise profiles as considered as ontologies. This article presents a framework based on the ontological and context awareness during the task integration and matching in order to form collaborative processes in the manner of cross enterprise collaboration.

In recent years, with the growth of the Internet technology, the users of virtual community not only play the role of the information receiver but also a very important one to provide information. However, there is large amount of information aggregated daily and therefore information overloading has become a very serious problem. Under this situation, how to find information efficiently is also a very important issue. In this paper, we believe users in a virtual community may affect each other, especially those with high influence who have been called as Key Users. Therefore, we observe the biggest virtual community of movies on the Internet which is named IMDb (The Internet Movie Database). An architecture also has been proposed that combines Social Networks Analysis and the features of IMDb to discover those users who have high influence in the virtual community. We collected 17 months (January 2010 to May 2011) from IMDb including 17 366 users and 243 074 reviews. By applying the method we proposed, there are about 22 key users and 111 reviews were discovered. We also use the box office of the movies to justify our results.

Online social media have been playing an important role of creating and diffusing information to many users. It means the users can get cognitive influence to the other users. Thus, it is important to understand how the information can be diffused by interactions among users through online social media. In this paper, we design a social media monitoring system (called ``TweetPulse'') which can analyze and show meaningful diffusion patterns (DP) among the users. Particularly, TweetPulse focuses on visualizing information diffusion in Twitter, given a certain time duration. Also, this work has investigated the relationships 1) between DP and event detecting, 2) between DP and emotional words, and 3) between DP and the number of followers of the users. Thereby, to understand the continuous patterns of the information diffusion, we propose two different types of analytic methods, which are 1) macroscopic approach and 2) microscopic approach. For evaluating the proposed method, we have collected and preprocessed the dataset during about 4 months (14 March 2012 to 12 July 2012). As a conclusion, TweetPulse has helped users to easily understand DP from a large scale dataset streaming through Twitter.

Incremental data mining has been discussed widely in recent years, as it has many practical applications, and various incremental mining algorithms have been proposed. Hong et al. proposed an efficient incremental mining algorithm for handling newly inserted transactions by using the concept of pre-large itemsets. The algorithm aimed to reduce the need to rescan the original database and also cut maintenance costs. Recently, Lin et al. proposed the Pre-FUFP algorithm to handle new transactions more efficiently, and make it easier to update the FP-tree. However, frequent itemsets must be mined from the FP-growth algorithm. In this paper, we propose a Pre-FUT algorithm (Fast-Update algorithm using the Trie data structure and the concept of pre-large itemsets), which not only builds and updates the trie structure when new transactions are inserted, but also mines all the frequent itemsets easily from the tree. Experimental results show the good performance of the proposed algorithm.

Many real world problems deal with ordering of objects instead of classifying objects, although most of research in data analysis has been focused on the latter. One of the extensions of classical rough sets to take into account the ordering properties is dominance-based rough sets approach which is mainly based on substitution of the indiscernibility relation by a dominance relation. In this paper, we address knowledge measures and reduction in incomplete fuzzy information system using the approach. Firstly, new definitions of knowledge granulation and rough entropy are given, and some important properties of them are investigated. Then, dominance matrix about the measures knowledge granulation and rough entropy is obtained, which could be used to eliminate the redundant attributes in incomplete fuzzy information system. Lastly, a matrix algorithm for knowledge reduction is proposed. An example illustrates the validity of this method and shows the method is applicable to complex fuzzy system. Experiments are also made to show the performance of the newly proposed algorithm.

In this paper we compare usefulness of statistical techniques of dimensionality reduction for improving clustering of documents in Polish. We start with partitional and agglomerative algorithms applied to Vector Space Model. Then we investigate two transformations: Latent Semantic Analysis and Probabilistic Latent Semantic Analysis. The obtained results showed advantage of Latent Semantic Analysis technique over probabilistic model. We also analyse time and memory consumption aspects of these transformations and present runtime details for IBM BladeCenter HS21 machine.

In comparison to dense matrices multiplication, sparse matrices multiplication real performance for CPU is roughly 5–100 times lower when expressed in GFLOPs. For sparse matrices, microprocessors spend most of the time on comparing matrices indices rather than performing floating-point multiply and add operations. For 16-bit integer operations, like indices comparisons, computational power of the FPGA significantly surpasses that of CPU. Consequently, this paper presents a novel theoretical study how matrices sparsity factor influences the indices comparison to floating-point operation workload ratio. As a result, a novel FPGAs architecture for sparse matrix-matrix multiplication is presented for which indices comparison and floating-point operations are separated. We also verified our idea in practice, and the initial implementations results are very promising. To further decrease hardware resources required by the floating-point multiplier, a reduced width multiplication is proposed in the case when IEEE-754 standard compliance is not required.

Visual odometry estimates a robot's ego-motion with cameras installed on itself. With the advantages brought by camera being a sensor, visual odometry has been widely adopted in robotics and navigation fields. Drift (or error accumulation) from relative motion concatenation is an intrinsic problem of visual odometry in long-range navigation, as visual odometry is a sensor based on relative measurements. General error analysis using ``mean'' and ``covariance'' of positional error in each axis is not fully capable to describe the behavior of drift. Moreover, no theoretic drift analysis is available for performance evaluation and algorithms comparison. Drift distribution is established in the paper, as a function of the covariance matrix from positional error propagation model. To validate the drift model, experiment with a specific setting is conducted.

Medical simulation is a new method of education in medicine. It allows training medical students or practitioners without the need to involve patients and makes them familiar with various kinds of examinations, especially related to medical imaging. Simulators that visualize examinations or operations require large computing power to keep time constraints of output presentation. A common approach to this problem is to use graphics processing units (GPU), but the code is not portable. The method of parallelization of processing is more important in component environments, to allow calculating projections in real time. In this paper parallelization issues in the ultrasound view simulation based on provided computer tomography images are analyzed. The proposed domain decomposition for this problem leads to significant reduction in simulation time and allows obtaining an animated visualization for currently available personal computers with multicore processors. The use of a component environment makes the solution portable and makes it possible to implement a web-based application that is the basis for eTraining. The method for creating animation in real time for such solutions is also analyzed.