memo / msa_bib_title_gen

Generate fun titles (or phrases) that sound scientific by un-abbreviating a given word using titles from a bibtex file

Generate fun titles (or phrases) that sound scientific by un-abbreviating a given word using titles from a bibtex file.

Useful for generating fun titles for papers or other shennigans.

e.g.

JUNKHEAD:

• Joint Unconstrained Networks Kinematic Handwriting Evaluating Artistic Density
• Joint Unreasonable Neocortex Knowledge Hierarchical Ecosystems Agent Detection
• Joint Understanding Novel Kinematic Hand Embeddings Arm Database
• Joint Unsupervised Neural Kinematics Human Evaluation Algorithmic Design

BUTTFACE:

• Bayesian Unique Theorem Taxonomy Formal Autonomous Convolutional Emotion
• Biological Unsupervised Transfer Tracking Fictional Algorithmic Coordination Experience
• Basic Unconstrained Trajectory Tutorial Fitting Animation Calligraphic Embedding
• Backpropagation Unreasonable Temporal Translation Framework Analysis Connectionist Environment

(results aren't always amazing, manual mix-and-matching from multiple samples gives best results)

#USAGE:

Basic usage:

msa_bib_title_gen --phrase [phrase to unabbreviate] --bib_path [path to bibtex file]

In case you don't have a bibtex file, I've provided mine (msa.bib). Papers on mostly AI, philosophy, cognitive science, robotics, graphics etc.

usage: msa_bib_title_gen.py [-h] [--phrase PHRASE] [--bib_path BIB_PATH] [--count_thresh COUNT_THRESH] [--sample_count SAMPLE_COUNT]

optional arguments: -h, --help : show this help message and exit --phrase PHRASE : phrase to unabbreviate --bib_path BIB_PATH : path to bibtex file --count_thresh COUNT_THRESH : ignore words occuring less than this many times --sample_count SAMPLE_COUNT : how many to sample

#REQUIREMENTS:

• python 2.7 (probably runs on 3.x too, maybe with minimal changes)
• numpy
• bibtexparser

#TODO:

• add support for text files (instead of bibtex)
• add support for arxiv (scrape arxiv for titles, by category, CS, Physics, Astronomy etc)
• error handling
• use end probability
• make online version