Contributions published at Business Intelligence (Daning Hu)

null

Analysing movie data and their box-office information, runtime, MPAA based on top 50 or top 250 movies is an often-discussed topic in research. This work did a research based real-life movie data set where the data set was extended with the Internet Movie Database (IMDb) movie information that build the basement for the data analysis, with the focus on genre information and genre clustering. Based on the different cluster analyses we identified eight clusters: action, thriller, crime ñ comedy, family, adventure, animation, fantasy ñ drama ñ action adventure ñ drama, war, historyóscifi, thriller, action ñ comedy ñ biography, drama.

Efficiently assigning human resources in Workflow Management Systems (WfMS) is a vital aspect when implementing them in corporate environments. This thesis expands on existing work in the field of role resolution in WfMS by extending the already researched Mixed Integer Linear Programming (MILP) methodologies and proposes a novel approach by introducing Reinforcement Learning (RL) based approaches. Both the extended MILP as well as the RL based methods outperform the traditional approaches up to a 1.3-fold speedup. RL based methods lay the foundations for extensions by using alternative methods such as Inverse RL (IRL) and Apprenticeship Learning (AL). Future work could reconcile traditional MILP and AL based methods by using the former as the "expert" agent performing role resolution and the latter observing its behavior in order to learn from it.

Smog disasters are becoming more and more frequent and may cause severe consequences on the environment and public health, especially in urban areas. Social media as a real-time urban data source has become an increasingly effective channel to observe people׳s reactions on smog-related health hazard. It can be used to capture possible smog-related public health disasters in its early stage. We then propose a predictive analytic approach that utilizes both social media and physical sensor data to forecast the next day smog-related health hazard. First, we model smog-related health hazards and smog severity through mining raw microblogging text and network information diffusion data. Second, we developed an artificial neural network (ANN)-based model to forecast smog-related health hazard with the current health hazard and smog severity observations. We evaluate the performance of the approach with other alternative machine learning methods. To the best of our knowledge, we are the first to integrate social media and physical sensor data for smog-related health hazard forecasting. The empirical findings can help researchers to better understand the non-linear relationships between the current smog observations and the next day health hazard. In addition, this forecasting approach can provide decision support for smog-related health hazard management through functions like early warning.

null

null

null

Smog disaster is a type of air pollution event that negatively affects people's life and health. Forecasting smog disasters may largely reduce potential loss that they may cause. However, it is a great challenge since smog disasters are often caused by many complex factors. With the availability of huge amounts of data from the social web and physical sensors, covering information of air quality, meteorology, social event, human mobility, people's opinion, etc., it becomes possible to utilize such big data to forecast smog disasters. Especially, we can investigate the effect of social activities in smog disaster forecasting with the help of social web, which is ignored in traditional studies. In this paper, we propose a big data approach named B-Smog for smog disaster forecasting. It mainly has two components: 1) features extraction from multiple data sources to model the factors that indicate the appearance or disappearance of a smog disaster like traffic condition, human mobility, weather condition and air pollution transportation; 2) learning and predicting with heterogeneous features in multiple views. For the second component, we propose a prediction model based on an ensemble learning framework and artificial neural networks (ANNs), which achieves high accuracy in this application and can also be applied to other similar problems. We present the effectiveness of B-Smog through two cases studies in Beijing and Shanghai, and evaluate the accuracy of the prediction model through comparing it with some baselines. Moreover, the empirical findings of our study can also support decision making in smog disaster management.

Recent economic crises like the 2008 financial tsunami has demonstrated a critical need for better understanding of the topologies and various economic, social, and technical mechanisms of the increasingly interconnected global financial system. Such a system largely relies on the interconnectedness of various financial entities such as banks, firms, and investors through complex financial relationships such as interbank payment networks, investment relations, or supply chains. A network-based perspective or approach is needed to study various financial networks in order to improve or extend financial theories, as well as develop business applications. Moreover, with the advance of big data related technologies, and the availability of huge amounts of financial and economic network data, advanced computing technologies and data analytics that can comprehend such big data are also needed. We referred this approach as financial network analytics. We suggest that it will enable stakeholders better understand the network dynamics within the interconnected global financial system and help designing financial policies such as managing and monitoring banking systemic risk, as well as developing intelligent business applications like banking advisory systems. In this paper, we review the existing research about financial network analytics and then discuss its main research challenges from the economic, social, and technological perspectives.

null

null

When designing a business process, individual agents are assigned to perform tasks based on certain goals (i.e., designed process goals). However, based on their own interests, real-world agents often have different goals (i.e., agents’ goals) and thus may behave differently than designed, often resulting in reduced effectiveness or efficiencies of the executed process. Moreover, existing business process research lacks effective methods for discovering agents’ goals in the actual execution of the designed business processes. To address this problem, we propose an agent-oriented goal mining approach to modeling, discovering, and analyzing agents’ goals in executed business processes using historical event logs and domain data. To the best of our knowledge, our research is the first to adopt the agents’ goal perspective to study inconsistencies between the design and execution of business processes. Moreover, it also provides a useful tool for stakeholders to discover real-world agents’ actual goals and thus provides insights for improving the task assignment mechanism or business process design in general.

null

null

null

The stock market is a highly complex and interconnected system. Another level of complexity is added by the temporal evolution of the market. In this work, historic stock market data gathered from the New York Stock Exchange over a period of over 23 years will be analyzed. Specifically, a correlation analysis of price returns will be used to construct network representations of the market data. To capture the dynamics of the market situation different time window sizes and threshold values are applied to the data. The results show, that one optimal value can not satisfy all needs. A thorough understanding of the parameters is needed to select the appropriate variables for the specific task. Furthermore a novel approach to network formation for correlation networks over long time periods is proposed. This method allows the detection of weak but temporally persistent correlations in stock data, which are filtered out in the conventional method of network formation. This method could in future also provide insight in to other data sets of a dynamic nature.

Lately, we often heard about the Business Intelligence solutions but very less about the Process Improvement with the Business Intelligent. In order to continue, expand from other companies, Siemens Switzerland AG - Divison Mobility and Rail Systems (MOL / RL) would like to improve their processes and gain relevant information to make better and accurate decisions. As part of my master thesis, I analysed the MOL / RL processes and developed a BI solution by focusing in three defined use cases. The goals were to increase productivity, place the trend technology, gain in efficiency with faster access, and provide relevant information to make better and accurate decisions. My research findings show the following weaknesses: -“Inefficient process” -“No standards reports and different data sources” -“A lot of media disruption and manually work” -“No up-to-date information” -“No adequate technology” After those findings we have defined a uniform processes and based on it, we have developed BI solution with the Microsoft technology. This BI solution has been evaluated thereafter. In conclusion, based on this evaluation results with the new process definition and BI solution the goals from both side IT and business are achieved. Moreover the results of this research showed that with this new process definition and BI solution, the productivity can be increased more than 200% and very high efficiency can be gained.

The 2008 banking crisis demonstrated that there is a lack of effective methods for modeling and analyzing “exceptional but plausible” risk scenarios in bank stress testing. Existing stress testing practices mainly focus on modeling probability-based risk factors and events in banking systems using historical data. Rare (low probability) risk events that can cause financial crises in banking systems, such as the bankruptcy of Lehman Brothers, are largely ignored due to the lack of appropriate modeling and analysis methods. To address this problem, we propose an approach called Banking Event-driven Scenario-oriented Stress Testing (or simply, BESST) which has two main components: 1) an ontology-based event-driven scenario model (OESM), and 2) two analysis methods based on OESM for scenario recommendation and plausibility checking. The proposed BESST approach provides bank stress testing stakeholders an effective method for modeling and analyzing financial crisis scenarios that are rare but often have significant consequences.

In the problem management, there is a need for decision support. To fulfill this need, the IT service desk data takes a strategic role. This work analyzes how a categorization model may give the business access to decision support. On the one side, there is a need for automatic status classification of the data. On the other side, a flexible querying and reporting process for business users is required. Thus, this work analyzes a method how the categorization of service desk data can be modelled and maintained. The proposed method combines both status classification of structured features and text classification for the unstructured features. Although many classification methods have been tested in research, there is still a need for an automated classification process. Many supervised approaches have been found to be successful. However, they require training data. This work addresses these difficulties in the creation and maintenance of a knowledge base. To do so, a process is analyzed empowering the business to create and maintain a knowledge base for the classification. The approach is based on the Case-Based-Reasoning method. The developed design analyzes how business experts can be involved in the knowledge base creation and its maintenance. The categorization model is evaluated with a proof-of-concept. Results show how valuable service desk data become when made accessible to the business by the proposed model.