User Tools

Site Tools



This shows you the differences between two versions of the page.

Link to this comparison view

Next revision
Previous revision
projects:year5:16.10 [2019/08/21 10:24]
sally.johnson created
projects:year5:16.10 [2019/08/22 10:50] (current)
sally.johnson [Project - Documents]
Line 2: Line 2:
 ===== Project - Summary ===== ===== Project - Summary =====
 <WRAP leftalign box > <WRAP leftalign box >
-**Objectives:**+Many real-world systems are so large that capturing them entirely, analyzing them to extract information, and visualizing them for decision making are resource-consuming and challenging tasks. It is necessary to develop sampling and summarization approaches and integrate them with visualization interfaces to study large-scale graphs to understand the underlying real-world systems. We have designed and implemented a new similarity metric for use in graph summarization, allowing large graphs to be rendered more easily. The cross-platform, touch-based interface in Figure 1 illustrates the summarization hierarchy created by our metric with a real-world dataset and allows users to control the level of summarization.
-The overall objective of this research is to develop new methods for comparative knowledge discovery for ranking - specifically enabling decision makers to better compare objects through both new ranking methods and interactive visualization methods. The overall project has four specific objectives: +{{ :projects:year5:16.10_fig1.jpg?600 |}}
-  * Develop a scalable partial ordering based ranking method +
-  * Develop data-driven approaches (supervised and unsupervised) to learn to rank +
-  * Implement a web-based visualization and interaction methods for comparative knowledge discovery+
 </WRAP> </WRAP>
 ===== Project - Team ===== ===== Project - Team =====
-^ Team Member           ^ Role              ^ Email          ^ Phone Number   ^ Academic Site/IAB   +^ Team Member         ^ Role              ^ Email          ^ Phone Number   ^ Academic Site/IAB  
-Raju Gottumukkala     | PI                | Not available  | Not available  | UL Lafayette        | +Christoph Borst     | PI                | Not available  | Not available  | UL Lafayette       
-| Vijay Raghavan        | PI                | Not available  | Not available  | UL Lafayette        +| Mehmet Engin Tozal  | PI                | Not available  | Not available  | UL Lafayette       
-| Mehmet Tozal          | PI                | Not available  | Not available  | UL Lafayette        +Nicholas Lipari     | Graduate Student  | Not available  | Not available  | UL Lafayette       
-Moncef Gabbouj        | PI                | Not available  | Not available  | Tampere University +Md Enamul Haque     | Graduate Student  | Not available  | Not available  | UL Lafayette       |
-| Alexandros Iosifidis  | PI                | Not available  | Not available  | Tampere University +
-| Amirhossein Tavanaei  | Graduate Student  | Not available  | Not available  | UL Lafayette        +
-Siva Venna            | Graduate Student  | Not available  | Not available  | UL Lafayette        | +
-| Guanqun Cao           | Graduate Student  | Not available  | Not available  | Tampere University  |+
Line 25: Line 18:
 ===== Project - Impact and Uses/Benefits ===== ===== Project - Impact and Uses/Benefits =====
 <WRAP leftalign box > <WRAP leftalign box >
-Ranking is a very classical problem that is very relevant to each and every organization - as organizations rank departmentsproductsclientsor geographical regions for important decisionsYetranking - for example in situations like KPI dashboards use composite index with linear weighted models to compare objects; While existing visualization dashboards do provide drill down capabilities, they do not offer deeper understanding with respect to how well the objects compare with each other particularly conflicts or missing information. The proposed approaches offer clear benefits in terms of(1Reducing manual effort involved in developing weights for ranking by experts that would save time and money(2) Offers deeper insights into how objects are ordered for people to make better decisions in government and industry domains that could potentially save costs and improve the performance of decision making in the context of ranking.+Complex networks have been studied across different fields of science such as mathematicalscientificand engineering perspectives. For examplea citation network [15] is described as a set of academic papers where a connection represents one paper referencing anotherIn studying human disease propagationresearchers are interested in finding the correlation between diseases that are often diagnosed together. Diagnoses can form complex network with diseases connected based on the frequency of co-occurrence in a patient population. Another common example of complex networks can be the World Wide Web, the largest network humanity has ever built. In this network the nodes are documents and the links are the uniform resource locators (URLsembedded within the documents. 
 +The proper visualization and monitoring of business processes is crucial for any enterprise. By using weighted summarization, we can provide an integrated model of the overall process, reduce the size of the dataset, observe connectivity and relationships among business processesand identify what methods are suitable for analysis. The creation of graph summaries allows more efficient community detection within a variety of networks, such as product-consumer network and online social networks. 
 +For companies and business, it is crucial to retain customers, build loyal relationships, and thereby avoid customer acquisition costs. Having a graph of customers and modeling their behavior helps us tho have insight of the data and maximize customer loyalty, retention and lifetime value. Weighted summarization can assist in the analysis and prediction of a new product and its impact on customers and help us to find the core cluster of customers. By understanding the different clusters of customers, we can observe the patterns of firmographic, demographic, and behavioral trends that correlate strongly with high-volume, modestly valuable, and borderline churn customers. 
 +Due to the dynamics of the economy, the marketplace and company decision making can never be captured fully beforehand. However, by using a summarized model, we can estimate potential revenue sources (e.g., assist advertising agencies in choosing which companies to approach). We can identify which revenue source to pursue, what value to offer, how to price the value, and what markets can bear that prices. 
 +[15] J. Gehrke, P. Ginsparg and J. Kleinberg, "Overview of the 2003 KDD Cup," ACM SIGKDD 
 +Explorations Newsletter, vol. 5, no. 2, pp. 149-151, 2003.
 </WRAP> </WRAP>
Line 33: Line 35:
 ===== Project - Documents ===== ===== Project - Documents =====
projects/year5/16.10.1566401064.txt.gz · Last modified: 2019/08/21 10:24 by sally.johnson