Public | Automated Build

Last pushed: a year ago
Short Description
Explore associations between Microsoft abstracts and methods via text analysis (under development).
Full Description

Microfish

author collaborations

preview application

under development!
This project has only been worked on for an afternoon! Currently, the only view is to search and view the mean similarity score for a single author across all methods. More advanced visualizations, views for methods, and graphs will be coming soon.

Large File Storage

Important This repo uses Github Large File Storage (lfs) for some of the data! You may need to install and configure for the clone to work properly.

Running

Without Docker

  cd microfish
  python index.py

Then go to http://127.0.0.1:5000/ in your browser.

With Docker

  git clone http://www.github.com/vsoch/microfish
  cd microfish
  docker-compose up -d

To get the docker container IP address, you need to do:

  docker inspect microfish_web_1 |grep "IPAddress"

And it should be repeated twice (will have better solution for this). Then go to ${IPADDRESS}:5000 in your browser.

About the Visualization

I parsed ~5000 publications and associated authors from the Microsoft Research site, and mapped each abstract into a vector space derived from descriptions of methods. I could then calculate similarity of each paper to the methods themselves based on the context and words in the paper text and methods descriptions, respectively. While it would have been ideal to use the DOI associated with the papers to obtain full text, I only had an afternoon to put this together, and opted for this quicker method.

Docker Pull Command
Owner
vanessa
Source Repository

Comments (0)