PhD in Computer Science

nlp4twitter

Academic year 2015/2016

2. Some knowledge of probability, but the course will be self­contained. Perhaps many of the

participants have already read Jurafsky and Martin, that should be enough.

Sommario del corso

• Course objectives
• Learning assessment methods
• Program
• Suggested textbooks and readings
• Notes
• Teaching tools

Course objectives

In this tutorial, we will analyze some of the most recent natural language processing techniques for

extracting detailed information from tweets.

Learning assessment methods

1. A project, 80% of the final grade (preferably with some issues in Italian)

2. A presentation of an assigned paper, 20% of the final grade

Given that my Italian is non­existent, and that there's always much more work in English, we will try to

make this course a good contribution to Italian NLP.

Program

We will cover several topics in this course. Some of them include:

1. POS­Tagging

2. Sentiment analysis/classification

3. Polarity

4. Named­entity recognition

5. Event detection

6. topic identification/modelling

7. Latent semantic analysis

8. Dirichlect allocation

9. Language modelling

10. "interestingness"

11. ...

to name but a few of what we might cover. To this, we have to add the "high performance" part of it:

1. Frameworks: hadoop, spark

2. Parallel programming: parallel (bash), pyparallel (python)

This is simply a list of topics, I'd be very interesting in hearing what students might be interested in as

well. Shoot me an email!

http://leoferres.github.io/nlp.html

Note

The first class will be a tutorial on using the following tools:

1. Jupyter notebooks

2. NLTK & spaCY

3. Scikit learn

4. Matplotlib

and we will probably use some other tools in the Python stack.

• Teaching materials
• Exams and finals
• Go to Moodle
• Visit the forums