prepare Hu et al 2019 for reading group (10/11)

[BarbaraMcG]

• [x] 1. What is this paper about?
• [x] 2. Is it relevant to our project? If so, why and how?
• [x] 3. What could we use from this work in our project?
• [x] 4. Add some text about it to Overleaf
• [x] 5. Plan experiments (if appropriate)

The paper is this one: Hu, R., Li, S., & Liang, S. (2019). Diachronic Sense Modeling with Deep Contextualized Word Embeddings: An Ecological View, 3899–3908. https://doi.org/10.18653/v1/p19-1379


See https://github.com/Living-with-machines/HistoricalDictionaryExpansion/issues/6 for full list of papers .

[BarbaraMcG]

@kasparvonbeelen , feel free to use this template for "your" paper for tomorrow, if it's useful!

[BarbaraMcG]

1. What is this paper about?

This paper is about using dictionary data to learn sense embeddings, and applying this to word sense disambiguation (WSD) and detection of sense-level semantic change.

2. Is it relevant to our project? If so, why and how?

Yes, in various respects:

• it's about semantic change over time at the level of senses rather than words
• it uses dictionary data

Questions

It's not clear if it's OED = Oxford English Dictionary, diachronic (www.oed.com) or ODE = Oxford Dictionary of English, synchronic (https://www.lexico.com). However, from the examples they use, e.g. Table 1, and from their code page it seems they use ODE (https://www.lexico.com/definition/mean).

Possible challenges:

• finding and running the code: https://github.com/iris2hu/diachronic-sense-modeling This seems to be only about the visualisation tool
• OED senses are diachronic, so we may not have enough data. They select only words that appear at least 10 times a year for over 50 consecutive years.

3. What could we use from this work in our project?

• We could build sense embeddings from the OED and then perform WSD to disambiguate the tokens of machine in our historical corpus by matching them to OED senses.

4. Add some text about it to Overleaf

I've added some text in the Overleaf project, in the "Related work" section.

5. Plan experiments (if relevant)

To be discussed in the catch-up meeting on 10/11