|
|
|
|
Certain technology trends over the past decade are working to create a data rich society. First, dropping costs and ubiquitous deployment of devices such as sensors, RFID tags, video cameras, GPS devices, etc. is making our environment more instrumented than ever. Second, increased facilitation of 'connectedness technologies', such as the Internet, cell phones, PDAs, etc. is increasing the society's electronic interaction, most of which can be instrumented, and thus is being instrumented. Third, dropping costs and increasing capacity of storage devices has made it feasible to collect and store increasing amounts of data, and archive them for long periods of time. The convergence of these trends is causing the 'data glut' we observe around us today. This phenomenon is evident in personal and professional interactions, in both individual and organizational settings. Given the exponential rate of increase in 'data per capita' for the past decade, we have already reached a point where the collected data rarely, if ever, crosses the human eye. Furthermore, given the large volumes of data being collected, there is an ever increasing hope and expectation that it can reveal useful knowledge about the processes being monitored that is not known apriori. Thus, there is a need to develop techniques that can extract useful knowledge from this data in a (semi‑) automated manner. In an organizational context, this translates into an ability to extract useful metrics and models by analyzing data collected from various processes and interactions, and applying the knowledge gained for improved organizational/corporate functions. Our research is focused on two inter‑linked questions, namely
- How can existing technology be used to extract useful knowledge from data collected in various organizational processes; where the usefulness of knowledge comes from it enabling better understanding and/or optimization of some organizational function?
- What new technology needs to be developed to fulfill the gaps that exist in being able to answer the first question?
In this talk we show how latent organizational knowledge, of structures and processes, can be extracted using data mining techniques. This analysis illustrates the gap that often exists between formal organizational hierarchies and informal ones. We will also present a number of computational techniques, for data mining and warehousing, that we have developed, which are useful for carrying out such analyses. |
