/monolingual-word-aligner

Primary LanguagePython

A Word Aligner for English

This is a word aligner for English: given two English sentences, it aligns related words in the two sentences. It exploits the semantic and contextual similarities of the words to make alignment decisions.

Requirements

  1. Python NLTK
  2. The Python wrapper for Stanford CoreNLP

Installation and Usage

  1. Install the above tools.

  2. Change line 100 of corenlp.py, from "rel, left, right = map(lambda x: remove_id(x), split_entry)" to "rel, left, right = split_entry".

  3. Download the NLTK stopword corpus:

    python -m nltk.downloader stopwords

  4. Install jsonrpclib:

    sudo pip install jsonrpclib

  5. Download the aligner:

    git clone https://github.com/ma-sultan/monolingual-word-aligner.git

  6. Run the corenlp.py script to launch the server:

    python corenlp.py

  7. To view the aligner in action, run testAlign.py. (Word indexing starts at 1.)