Overview / Usage

The DeepCare chatbot is capable of learning to answer customer questions. Using a hybrid approach of NLP and Deep Learning, it tries to combat logical fallacies that occur in pure deep learning bots, while still coming up with unique answers.

In particular, it uses a sequence-to-sequence (seq2seq) long-short-term-memory LSTM deep learning model to capture intricacies in questions. As organisations cannot afford a bot making logical mistakes, verification through NLP is used. This two-step model prevents the downside of "no control" on deep learning, as well as the too static nature of classical rule based NLP models, and thus enables potentially higher quality answers.

Methodology / Approach

It is using a combination of machine learning (tensorflow), Twitter data, Natural Language Processing using spaCy.
Using only Deep Learning is a limited approach to capture the complexity of a chatbot. Therefore, in both the pre- and postprocessing stages, traditional state-of-the-art NLP techniques are used to improve accuracy.

Technologies Used

• Intel Optimized Python
• Intel Optimized Tensorflow
• LSTM
• spaCy (using Cython)
• Dependency Tree