GithubHelp home page GithubHelp logo

mindis / acl2017-non-factoid-qa Goto Github PK

View Code? Open in Web Editor NEW

This project forked from ukplab/acl2017-non-factoid-qa

0.0 3.0 0.0 1.69 MB

Code for paper "End-to-End Non-Factoid Question Answering with an Interactive Visualization of Neural Attention Weights"

Python 61.67% Scala 17.66% TypeScript 6.47% CSS 5.26% JavaScript 1.10% HTML 7.83%

acl2017-non-factoid-qa's Introduction

End-to-End Non-Factoid Question Answering with an Interactive Visualization of Neural Attention Weights

This project contains multiple components that, together, build an end-to-end question answering system. A special emphasis lies on the visualization and interactive comparison of neural attention weights.

Please use the following citation:

@InProceedings{rueckle:2017:ACL,
  title = {End-to-End Non-Factoid Question Answering with an Interactive Visualization of Neural Attention Weights},
  author = {R{\"u}ckl{\'e}, Andreas and Gurevych, Iryna},
  publisher = {Association for Computational Linguistics},
  booktitle = {Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (ACL 2017)},
  pages = {(to appear)},
  month = aug,
  year = {2017},
  location = {Vancouver, Canada}
}

Abstract: Advanced attention mechanisms are an important part of successful neural network approaches for non-factoid answer selection because they allow the models to focus on few important segments within rather long answer texts. Analyzing attention mechanisms is thus crucial for understanding strengths and weaknesses of particular models. We present an extensible, highly modular service architecture that enables the transformation of neural network models for non-factoid answer selection into fully featured end-to-end question answering systems. The primary objective of our system is to enable researchers a way to interactively explore and compare attention-based neural networks for answer selection. Our interactive user interface helps researchers to better understand the capabilities of the different approaches and can aid qualitative analyses.

Contact person: Andreas Rücklé

https://www.ukp.tu-darmstadt.de/

https://www.tu-darmstadt.de/

Don't hesitate to send us an e-mail or report an issue, if something is broken (and it shouldn't be) or if you have further questions.

This repository contains experimental software and is published for the sole purpose of giving additional background details on the respective publication.

Project structure

This repository contains the three sub-projects as described in the related paper:

  • Candidate-Retrieval: Indexes all answers of a data source and allows to find answers with lexical overlap to a question
  • Candidate-Ranking: Contains a framework to train neural network models for answer-selection. Contains a webservice that allows to re-rank answers in regard to their relevance to a given question.
  • QA-Frontend: The webfrontend that allows the user to ask questions and retrieve answers. All results will contain neural attention weights which researchers can view, inspect, and compare.

Each individual project has its own installation instructions in separate readme files. It is also possible to install each service on a separate machine, because the services communicate over HTTP.

To setup the project, it makes sense to first install and run the Candidate-Retrieval application, afterwards setup and train a simple neural network with the Candidate-Ranking, and finally install the QA-Frontend to pull everything together.

API Documentation

QA-Frontend uses the Candidate-Retrieval and Candidate-Ranking services to find and rank relevant answers. Both services provide a public HTTP REST APIs, which are documented in API.md. Researchers can easily substitute individual service implementations with their own approaches.

Screenshots

Answer Ranking

It is possible to query any question text and inspect the resulting neural attention weights in the question as well as the answer.

Screenshot

Side-By-Side Comparison

If multiple candidate ranking instances with different models are running, it is possible to compare the neural attention weights of both approaches side-by-side within the same window.

Screenshot

acl2017-non-factoid-qa's People

Contributors

arueckle avatar

Watchers

 avatar  avatar  avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.