Slide Ten - Machine Learning - Clustering

To generate what Professor Eastman calls aggregations of functional entities, we need to cluster the words which appear in the same context because IR theory has shown that words close in context have a shared dependency. Since context similarity is based on the spread of data in the matrix, I use the K-Means algorithm, a common clustering technique based on the spread of data. The clustering results show for example than discussion of "design" is strongly associated with the "controller" for example, or that the documents describe the "control" of the "system." Similarly, the "transfer function" of the system is related to the "response" and the "error" of the system.