Research Project: Word Level Techniques for Historical Spelling Normalization
Why did you apply for this internship?
I wanted to get experience doing real research software engineering, and the opportunity to learn from experienced researchers. I would like more experience in research, as I find it interesting and would be interested in pursuing research in the future. I am also interested in this project specifically, as I like the interdisciplinary nature of it, as its applications could be useful in other fields outside of computer science.
What did you hope to gain in completing this project?
I hoped to gain a better understanding of how academic research in this field is really conducted, and improve my programming and research skills.
Project Overview
Produced a framework that allows easy development and testing of historical spelling normalisation techniques, and makes it easy to switch between different techniques and data.
Using this framework to test word level techniques for normalising historical spelling.
What were the key results of your research project?
- Developed a framework to test and develop historical spelling normalisation techniques
- Tested various word-level techniques
I found that the word level techniques tested (including statistical and neural translation models, a Levenstein distance-based approach, and others) were less effective than the state-of-the-art character-level methods.
The most effective word-based technique tested was a simple method that learned mappings from historical spellings to modern spellings using aligned, normalised, and non-normalised corpora, and this outperformed the more complex techniques.
How do you feel you have benefited from completing this internship, and has it made you consider future career paths?
The biggest benefit I have gained from this has been a major improvement in my programming skills. Working on a large project like this and being helped by more experienced programmers has helped develop my programming problem-solving skills and improved the cleanliness, readability, and efficiency of the code that I write.
It has definitely helped me consider a future in research software engineering, as I have enjoyed working on a research project like this.
Download presentation slides: