Description

Sentence_classification project aims to create python module with easy to use Tensorflow based classifiers for text classification on the level of single line.
Enron manually annotated database was used as a base for all evaluations.
Two approaches of classification are implemented now:

• MovingWindowSentenceClassifier - sliding window, with odd vertical dimension - classifier bsaed on CNN/BiGRU layers,

• self-created embeddings version summary, diagram

• ContextBranchSentenceClassifier - context branches, with preceeding and following lines passed to separated branches,

• self-created embeddings version summary, diagram

Directories content:

• classes - ready to use classifiers
• dataset - enron files annotated
• examples - example scripts and Jupyter notebooks
• optim - results of crossvalidated hyperparameters optimization using Neural Network Intelligence
• test - unittests