Exercise 1

• Run exercise-1.py file in src/exercises/;
• It applies system represented by the image located in mod dir, with an example document in file system.

Exercise 2

• Run exercise-2.py file in src/exercises/;
• It applies the same system but with variants trying to improve results;
• The results from exercise-1 and this exercise are represented in a txt file localed in results dir and it uses a dataset with 1500 abstract documents.

Exercise 3

• Run exercise-3.py file in src/exercises/
• It applies another approach to key phrases extraction: rank aggregation

Exercise 4

• Tornado package is necessary;
• Conection to internet is necessary to get RSS file;
• Run exercise-4.py in src/exercises/ to start web server;
• Open web browser and do a request to "localhost:8888" or if you changed the port "localhost:newPort"
• This request return an html page which allows to see every sport article and its key phrases. Also, it has filter mechanisms.

Extra files

• During the development, we generated a pickle file for word vectors that exceed the max unique file size. The big_files_url contains an URL to download that file. It should be saved into src/files/ and the name must be preserved or you might change on code.