PAN2018: PHAN Style Change Detection

A Parallel Hierarchical Attention Network for Style Change Detection

Authors : Marjan Hosseinia and Arjun Mukherjee

We propose a model for the new problem of style change detection. Given a document, we verify if it contains at least one style change. In other words, the task is to investigate whether it is written by one or multiple authors. The model is composed of two parallel attention networks. Unlike the conventional recurrent neural networks that use the character or word sequences to learn the underlying language model of documents, our model focuses on the hierarchical structure of the language and observes the parse tree features of a sentence using a pre-trained statistical parser. Besides, our model is independent of style change positions although they are given during the training phase. The reason is to have a more applicable approach to the real world problems where such information is not available. PAN 2018 results show that it achieves 82% accuracy and stays at the second rank.

prerequisits:

Stanford CoreNLP, pytorch 0.4, nltk, tqdm, pycorenlp

Dataset:

PAN 2018 Style Change Detection. The csv files of the dataset can be found in data folder.

Stanford CoreNLP:

Dowanload Stanford CoreNLP

Unzip the file:

Unzip stanford-corenlp-full-2018-02-27.zip

Run the server:

cd stanford-corenlp-full-2018-02-27

java -mx4g -cp "*" edu.stanford.nlp.pipeline.StanfordCoreNLPServer -port 9000 -timeout 15000

train:

sypt_train.py -c csv_files_path -o ptf_files_path

csv_file_path: path to the csv train and validation files

ptf_files_path: path to ptf_train and ptf_validation files

e.g. sypt_train.py -c data/ -o data/

test:

sypt_test.py -c csv_files_path -o ptf_files_path

csv_file_path: path to the csv test file

ptf_files_path: path to ptf_train and ptf_test files and the json output file

e.g. sypt_test.py -c data/ -o data/

The attention implementation has been inspired from this code.