search

a1da4 / paper-survey

Summary of machine learning papers
32 stars 0 forks source link

Reading: A bottom up approach to category mapping and meaning change #228

Open a1da4 opened 2 years ago

a1da4 commented 2 years ago

0. Paper

1. What is it?

They investigated the relationship between word sense clusters and semantic change.

2. What is amazing compared to previous works?

Previous works are based on isolated case studies, but this paper handles all words.

3. Where is the key to technologies and techniques?

  1. train time-specific word vectors over time
  2. in each time period, cluster word vectors
  3. calculate the correlation between
    • distances from centroids (e.g. vec(apple)_1960s and centroid of its cluster)
    • likelihoods of semantic change (e.g. vec(apple)_1960s and vec(apple)_1970s)

4. How did evaluate it?

スクリーンショット 2022-02-23 3 38 50

5. Is there a discussion?

6. Which paper should read next?