Dechrissen / buzzrec

Recommendation engine for LingBuzz papers

buzzrec

A content-based filtering recommendation engine for academic papers in linguistics from LingBuzz.

buzzrec can be downloaded and used locally as a command-line tool.

Table of Contents

1. Requirements
2. Installation
3. Usage

Requirements

• Python 3
• numpy==1.19.2
• rake_nltk==1.0.4
• pandas==1.1.4
• requests==2.24.0
• nltk==3.5
• beautifulsoup4==4.9.3
• scikit_learn==0.23.2

Installation

First, download this repository with the green Code button, or via git like below:

$ git clone https://github.com/Dechrissen/buzzrec.git

Next, install project dependencies. cd to the buzzrec directory, then:

$ pip install -r requirements.txt

Usage

Initial setup

Populate config.json with the following information:

• keywords: keywords that define your interests

For example:

{
  "keywords" : ["computational phonology", "context free grammars", "french vowels"]
}

A note on keywords: Try to make your keywords more specific than 'phonology' or 'syntax', otherwise the initial data collection will take a while. Each keyword will make a new query to LingBuzz; the narrower the term, the more specific the results.

Using the tool

To run the tool, cd to the buzzrec directory, then:

$ python recommender.py

The 10 most recent LingBuzz paper uploads will be compared against your specific tastes, and the most similar paper will be recommended to you along with a link to its PDF. You will also see a new file user.csv in the project directory. This acts as your user model.

Starting over (deleting a user model)

To have buzzrec re-create your user model according to new keywords, simply delete user.csv and update config.json with new keywords before running the tool. Otherwise, your user model will be saved for repeated use.