https://github.com/tbarlow12/wiki-answer

I wanted to create a question answerer for Wikipedia articles. This project reads articles and responds to a set of questions, getting 31% accuracy on the test set of questions
https://github.com/tbarlow12/wiki-answer

nlp python question-answering spacy wikipedia

Last synced: about 2 months ago
JSON representation

I wanted to create a question answerer for Wikipedia articles. This project reads articles and responds to a set of questions, getting 31% accuracy on the test set of questions

• Host: GitHub
• URL: https://github.com/tbarlow12/wiki-answer
• Owner: tbarlow12
• Created: 2017-01-12T02:12:00.000Z (over 8 years ago)
• Default Branch: master
• Last Pushed: 2017-05-16T12:11:36.000Z (almost 8 years ago)
• Last Synced: 2024-10-19T17:33:07.762Z (7 months ago)
• Topics: nlp, python, question-answering, spacy, wikipedia
• Language: HTML
• Homepage:
• Size: 59.5 MB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.txt

Awesome Lists containing this project

README

        
This is my question-answering project for Information Extraction

This folder contains the source code that I wrote, as well as the datasets I used, but it leverages a few different external libraries, as documented in my technical design

--MY SOURCE CODE--

    /helpers - This is a Python module I created with helper functions required to support question answering.

    These helpers could be anything from reading a file to searching for RegEx matches to finding the Jaccard similarity of two sentences

    /questionAnswering - This is the Python module with the Question Answering algorithm I used. As mentioned above,

    I tried to encapsulate any methods/information non-essential to the algorithm inside the /helpers module and use it here

    test.py - Script to actually run the test for the accuracy of the question-answerer

--DATASETS--

    /datasets

        /train - Two datasets used in development (S08 and S09)

        /test - One dataset used only for testing (S10)

--RUNNING THE PROGRAM--

At runtime, the script fetches those libraries and uses them in the algorithm

To run the program with the development sets, type the following command into a CADE lab Linux terminal:

sh qa.sh

To run the program with the test set, type the following command into a CADE lab Linux terminal:

sh full.sh

I have tested to be sure that permissions are open for the necessary folders.