This master thesis discusses two main tasks of computational etymology. First, finding cognates in multilingual text. Second, finding underlying correspondence rules by aligning cognates.
For the first part, I briefly described two categories of methods in identifying cognates: symbol based and phonetic based. For the second part, I described the Etymon project, which I had been working in. The Etymon project uses a probabilistic method and Minimum Description Length principle to align cognate sets. The objective of this project is to build a model which can automatically find as much information in the cognates as possible without linguistic knowledge as well as find genetic relationship between those languages. I also discussed the experiment that I did to explore the uncertainty in the data source.
