Skip to content

Hierarchical LSTM Network for Sentence Level Classification

Notifications You must be signed in to change notification settings

CT15/Hierarchical-LSTM

Repository files navigation

Hierarchical LSTM

The model was initially implemented to solve multiple interventions problems for MOOC discussion forums. This is an improvement to the simpler BoW-LSTM model.

MOOC

MOOC stands for Massive Open Online Courses. It is an online course aimed at unlimited participation and open access via the web.

Source: Wikipedia

Some examples of MOOCs include Coursera and edX.

Multiple Intervention Problems

Given a discussion forum thread containing a list of posts ordered from the earliest timestamp to the latest timestamp (Note: This is one way to look at the problem. See postia for further details), determine which posts in this thread require/merit instructor interventions. It is a binary classification problem.

Instructor intervention is a term used to refer to instructor's reply to a particular post.

Model Architecture

Architecture

Additional Requirement

This model uses 300d pre-trained GloVe 6B embeddings. You can download it here.

About

Hierarchical LSTM Network for Sentence Level Classification

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages