cluster_outlier

This repository contains the code and embeddings for Outlier Detection and Word Similarity estimation.

1. Outlier Detection:

• The script to run the task is outlier_detection/prototype.py and it can be run on the Camacho-Collados_Dataset.

• A preprocessed version of the Camacho-Collados_Dataset can be found in preprocessed_datasets/Camacho-Collados_Dataset.txt.

• Embeddings are filtered to contain only the required vectors, and they are provided in embeddings/

• The basic command to run the script:

     python prototype.py --input $DIR_DATA --embedding $DIR_EMBEDDING
  

• To learn about the options provided by the script, run the following command:

    python prototype.py --help
  

• Tip : With minor changes, the script can be run on a preprocessed version of Blair datasets, also provided in the folder preprocessed_datasets/. The script requires to accomodate the varying number of items in each Blair cluster.

2. Word Similarity:

• The script to run the task is word_sim/wordsim.py and it can be run on the MEN, SimLEX, and WordSIM353 datasets.

• The command to run the script:

     python wordsim.py --input $DIR_DATA --embedding $DIR_EMBEDDING
  

• To learn about the options provided by the script, run the following command:

     python wordsim.py --help