Skip to content

Latest commit

 

History

History
71 lines (37 loc) · 2.21 KB

README.md

File metadata and controls

71 lines (37 loc) · 2.21 KB
Raw

aps

(for « Analyse de papiers scientifiques ») Analysis of scientific papers and results visualization.

Project done during our 1st year at Télécom ParisTech, in June 2016. We mined scientific publications from our school to generate insights about the topics and the authors.

For instance we built visualization about the various topics covered by the papers, and built charts to highlight links between authors.

Plugins used:

  • nltk for natural language processing

  • latexcodec to work with LaTeX

  • Flask to code the server

  • pdfMiner to extract text from pdf

  • bibtexparser to extract data from bibtex files

  • rdflib to work with RDF and represent information

  • pyspotlight to connect to DBpedia and retrieve semantic categories/topics

  • datetime for time computations

To install them all (Linux): → sudo pip install nltk latexcodec Flask pdfMiner bibtexparser rdflib pyspotlight dateutil

How to view the wordcloud of topics:

  1. You will need a command line inside the source folder. You can execute cd aps/ after cloning our repository.

  2. Launch python src/server.py to launch the Flask server.

  3. Go to http://localhost:5000/, this is the index page

  4. Launch the computation by going to http://localhost:5000/init_wordcloud. You can follow the computations in the console.

  5. When the page is done loading, it will print something like Word Cloud for biblio.bib done. you can check the word cloud on the index page.

  6. Enjoy!

How to view the coauthoring graph:

  1. Make sure you have telecom.bib in your aps/data folder. (If you don't, you can put another .bib file and rename it)

  2. You will need a command line inside the source folder. You can execute cd aps/ after cloning our repository.

  3. Launch python src/main_graph.py to launch the computations.

  4. Launch python src/server.py to start the server if the server isn't already on. If it's already on, there's no point restarting it as it refreshes.

  5. Go to http://localhost:5000/, this is the index page

  6. Enjoy!

Contributors:

Loïc Herbelot

Antoine Sueur

Romeo Brofiga

Adrien Lagasse