Dr. Nematollaah Shiri

Abstract

In this talk we will review major steps to transform data management to management data. This has implications for and impacts on a wide range of different research and development opportunities and activities. In addition to representing and processing date in most applications, we also need abilities to analyze, reason, and explain the data (ARE, pronounced as "areh"). Such capabilities are gaining popularity and essential too in particular when the data is large and/or complex. ARE focuses on methods to discover hidden information (jewel) in data (dirt),  and/or explaining query answers. For large organizations and industries, ARE is essential for planning, management, decision making.

To academia, ARE provides new research and training opportunities, which for industry, can result in activities for development of more efficient and scalable tools and techniques.

Abstract

Health care systems worldwide collect an enormous amount of data in electronic health records and other primary and secondary healthcare information systems. In recent years, there has been an increased focus on making better use of the collected data for monitoring, prevention, diagnosis, treatment, prognosis, administration, etc.

Data-driven health technology is interdisciplinary in nature requiring expertise from various fields such as health technology, computational intelligence, and health science. There is a huge potential, but also prominent challenges when developing and deploying novel technical solutions to the benefit of patients (and their relatives) and health professionals.

The talk will discuss various aspects of data-driven health technology and provide examples of novel intelligent systems based on health care data.

Abstract

Finding similarity between the semantic units in a text is a crucial issue which has a wide range of applications in text mining and text interpretation. A focal concern in this regard is extracting the key points from a text . Within this scope it would be important to figure out (i) in what way two semantic units could be regarded similar (or relevant) to each other, and (ii) how the shared aspects between two semantic units could be extracted. Finding the shared aspects between semantic units can in turn lead to conceptualizing the similarity between them.This matter is quite significant in the cases where a text is complex in nature from the viewpoint of both the included propositions as well as the variation of the standpoints employed in them.

A variety of approaches exist for conceptualizing similarity between two semantic units out of which (i) assessing similarity based on local similarities, (ii) making connection between two semantic units through activating a number of operators, and (iii) investigating a semantic unit from the viewpoint of another semantic unit based on the related propositions, can be mentioned. It is obvious that, in the third approach, the principle of having direct similarity is not necessarily consistent, and instead it would be crucial to show how through linking the corresponding propositions a novel semantic entity that in turn holds some utility in a particular context, can be established.

In this speech, having had a brief review on the existing approaches to finding similarity between the semantic units in a text, two cognitive frameworks will be discussed for conceptualizing similarity as an approach to extracting the key points. The first framework is based on activating a number of operators which are significant from either a linguistic or a psychological viewpoint, and the second framework is based on the idea of analyzing a semantic unit from the viewpoint of another unit to show to what extent the two units can be compatible with each other. Finally, some examples will be presented with regard to these frameworks .and their roles in realizing the possible kinds of similarity between different kinds of semantic units.