Time filter

Source Type

Lin W.,Huazhong University of Science and Technology | Yu D.Y.,Huazhong University of Science and Technology | Zhang C.,Huazhong University of Science and Technology | Liu X.,University of Michigan | And 4 more authors.
Journal of Cleaner Production

Industry is responsible for nearly half of the global energy consumption. Recent studies on sustainable manufacturing focused on energy saving to reduce the unit production cost and environmental impacts. Besides energy consumption, certain manufacturing activities in machine shops, such as the use of cutting fluids, disposal of worn tools, and material consumption, also cause other environmental impacts. Since all these activities lead to carbon footprint directly or indirectly, carbon footprint can be employed as a new and overall environment criterion in manufacturing. In this study, an integrated model for processing parameter optimization and flow-shop scheduling was developed. Objectives to minimize both makespan and carbon footprint were considered simultaneously, which was solved by a multi-objective teaching. -learning-based optimization algorithm. Furthermore, three carbon-footprint-reduction strategies were employed to optimize the scheduling results: (i) postponing strategy, (ii) setup strategy, and (iii) processing parameter preliminary optimization strategy. In the theoretical aspect, the strategies greatly improved the performance of the optimization results through reducing machine idle time and cutting down the search space. From the perspective of practical applications, these strategies greatly help elevate production efficiency and reduce environmental impacts. © 2015 Elsevier Ltd. Source

Sun C.,National School in Computer Science | Li Q.,National School in Computer Science | Cui L.,National School in Computer Science | Yan Z.,National School in Computer Science | And 2 more authors.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

The rapid growth of data makes it possible for us to study human behavior patterns. Knowing the patterns of human behavior is of great use to help us detect the unusual fraud human behavior. Existing fraud detection methods can be divided into two categories: pattern based and outlier detection based methods. However, because of the sparsity and complex granularity of big data, these methods have high false positive in fraud detection. In this paper, we propose an effective hybrid fraud detection method. We propose SSIsomap which improves isomap to cluster behaviors into behavior classes and propose SimLOF which improves LOF to conduct outlier detection, then we use Dempster-Shafer evidence Theory for combining behavior pattern evidence and outlier evidence, which yields a degree of belief of fraud to the new coming claim. The experiment result shows our method has significantly higher accuracy than exsiting methods in medical insurance fraud detection. © Springer International Publishing Switzerland 2015. Source

Zhu C.,Shandong University of Science and Technology | Li Q.,Shandong University of Science and Technology | Kong L.,Shandong University of Science and Technology | Wei S.,Shandong Hoteam Software Co.
Proceedings - 2015 12th Web Information System and Application Conference, WISA 2015

In big data epoch, one of the major challenges is the large volume of mixed structured and unstructured data, which comes in heterogeneous sources. Because of different form, structured and unstructured data are often considered apart from each other. However, they may speak about the same entities of the world. If a query involve both structured data and its unstructured counterpart, it is inefficient to execute it separately. The paper presents a novel index structure tailored towards the combinations of structured and unstructured data. The combined index is a joint index over structured database and unstructured document, based on entity co-occurrences. It is also a semantic index which describes the semantic relationships between entities and their multiple resources. We store the index as RDF graphs and queries are SPARQL-like. Experiments show that the associated index can not only provide apposite information but also execute queries efficiently. © 2015 IEEE. Source

Zhao J.,Shandong University of Science and Technology | Wang X.,Shandong University of Science and Technology | Yan Z.,Shandong University of Science and Technology | Wei S.,Shandong Hoteam Software Co.
Proceedings - 2015 12th Web Information System and Application Conference, WISA 2015

Event is a widely used concept these years. Many areas such as Natural Language Process, Information Retrieval have used event as the basic information unit in their research. So, the mining of event association is very necessary for our research. And it plays an important role business intelligence and researches of relations between events. Usually events are associated with others when they often occur in the vicinity of others or co-occur in the same context. However, there are some implicit associations we cannot mine only from sequence or context. In this paper, we aim to find associations of events under the background of Data Integration Systems. By using the structured information of data integration system, the background information of entities can be extracted to classify events. So we classify the events into different categories which makes it possible to mine the statistical information from event sequence. Furthermore, we generalize the association between event entities to predict the implicit association in our algorithm. We validate our method with experiments and results show the useful information in the area of business intelligence. © 2015 IEEE. Source

Zhang H.,Shandong University of Science and Technology | Yan Z.,Shandong University of Science and Technology | Sun C.,Shandong University of Science and Technology | Wei S.,Shandong Hoteam Software Co.
Proceedings - 2015 12th Web Information System and Application Conference, WISA 2015

As the sources of data become more and more diversified, the importance of data conflict detection is emerging. We are committed to research a new method, through the use of behavior pattern detection of heterogeneous data semantic conflict. We find that the structured data which can represents the behavior of an entity contradict from the reality behavior of the entity which can be got from unstructured text, which is often referred to as pattern conflict. So in this paper, we convert the structured data with semantic into data-converted event. Combine them with the text event extracted from unstructured text, according to the relation between entities, get a large event graph G. Find the common conflict pattern through frequent sub-graph discovery on graph G. Then use the common conflict patterns to detect conflict data. The experiment shows that our method can detect the conflict data effectively with a high recall. © 2015 IEEE. Source

Discover hidden collaborations