Computational Linguistics Reading Group

Hallo daar! In the reading group of the 18th of November, I will try to give an exhaustive presentation (but not a lengthy one!) of what my research project is about, showing you also some of the work done so far. The presentation should be divided into three main parts as it follows, although some adjustments could be made during this week:

1. The Resource: the CompWHoB Corpus (Esposito et al., 2015) is the corpus built and used in the research project. I will describe its main characteristics, focusing on the structural aspect and on the NLP pipeline used to process it.
2. Research Area/Methodology: the methodology employed in this work is the Distributional Semantics. As everybody probably knows, it is based on Zellig Harris’ distributional hypothesis (Harris, 1954). More specifically, in this work the Distributional Semantic Model known as Temporal Random Indexing (Basile et al., 2014) is used. I will talk about how this DSM allows to analyse word meaning change over time and how it can represent an important tool of analysis for both social scientists and linguists (Esposito et al., forthcoming).
3. Work done so far & Aims of the project: I would like to pay some attention to a topic modelling task carried out in a forthcoming paper (Esposito et al., forthcoming) where the “classical” approach based on the latent Dirichlet Allocation (LDA) has been compared to a framework employing word embeddings generated by Word2Vec model (Mikolov & Dean, 2013). Finally, I will discuss what the final aim of my project is.

I know it sounds like a lot of stuff, but I promise not to take up too much of your time.

I hope to see the most of you this Friday as I look forward to hearing your valuable opinions.

See you on Friday. Until then, have a great week!

Fabrizio