Concept Sequence Tagging for a Movie Domain

Second project of Language Understanding Systems, Spring Semester 2017, University of Trento.

The project consist in building a model for concept sequence tagging for a Movie Domain.

The task is described in the task file

The results are discussed in the report file.

Repository Structure

This repository contains:

• four folders, template, code, data, result;
• two .pdf, task and report; and
• two .md files, README and LICENCE

In the task pdf is reported the task of the assignment from the professor, in the report my explain of what I did, how and, with which results.

In the template folder is reported all the templates used for the testing of the algorithms

In the data folder is reported the dataset used. The dataset is divided in train and test and for each of them the .data file contains the tuple, (word concepts) and the .feats.txt file contains the tuples (word, Pos, Lemma). For a deeper analysis of the dataset composition please refer the first midterm project report

In the code folder is contained the main.py file, in which is contained the implementation of the algorithm discussed in the report.pdf; in the sandbox folder the main will dump all the working file; and the conlleval.pl script that allows the evaluation of the performances of the algorithm automatically.

Finally in the result folder will be written the performances.txt

Running the algorithm

The code/main.py can be launched as is without setting any parameters since a working configurations is already given by default.

cd code
chmod +x conlleval.pl
python2.7 main.py

Please note that the default configuration will clean up the sandbox when done.

For any further information about eh working of the algorithm do not hesitate to contact me or read the report.pdf