2nd Semester 2017/18: Distributional Semantics for Philosophy
If you are interested in this project, please contact the instructor by email.
Distributional semantics, the idea that the meaning of words can be derived from their linguistic context, has provided highly successful tools for computational linguists to solve all sorts of language engineering problems. Feeding a distributional semantic model with big data results in accurate models of meaning that provide reliable information on what words or groups of words have similar meanings.
Yet there is another field that has been thinking about “meaning” for a much longer time: philosophy. Can the distributional hypothesis of meaning, and its computational implementability provide any new insights in this field? On the one hand, philosophy is not characterized by big data but by tiny data. On the other hand, philosophical texts are often written in a highly precise style. Can we use distributional semantics as a meaning-measuring tool to help address philosophical research questions? This project is a way to get involved in this rather new field of research, and you will learn what is necessary to conduct a novel study of a philosophical meaning-related question using this new method.
As we do not expect everyone to have a background in philosophy, a corpus of digital philosophical texts and research questions from Arianna Betti’s e-Ideas project will be available.
In the first week (June 4-8), hands-on tutorial sessions will be organized introducing distributional semantics, Python-based tools for creating distributional semantic models, and tiny data methods for distributional semantics. In the second week (June 11-15) students will choose and present their research topic based on the tools, literature and data discussed so far. In the third week (June 18-22), the proposed experiments will be conducted, and the last week (June 25-29) will be reserved for writing up the studies as short papers.
The course will be co-organized by Aurelie Herbelot (Cimec, University of Trento).
Basic Python programming knowledge is expected. An interest in philosophy and in distributional semantics or corpus linguistics is helpful.
- Oral presentation of experiment proposal in week 2. To pass, students must also attend the other students’ presentations.
- Written report in the style of a short conference paper.
- Lenci, Alessandro. 2008. “Distributional Semantics in Linguistic and Cognitive Research.” 2008.http://linguistica.sns.it/RdL/20.1/ALenci.pdf.
- Herbelot, Aurélie, Eva Von Redecker, and Johanna Müller. 2012. “Distributional Techniques for Philosophical Enquiry.” In Proceedings of the 6th Workshop on Language Technology for Cultural Heritage, Social Sciences, and Humanities, 45–54. Association for Computational Linguistics. http://www.aclweb.org/anthology/W12-1008.
- Herbelot, Aurélie, and Marco Baroni. 2017. “High-Risk Learning: Acquiring New Word Vectors from Tiny Data.” Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, 304–9. http://aclweb.org/anthology/D17-1030.
More materials will be provided in the first session of the course.