Toy natural language understanding system. It takes example user requests (utterances) and tries to classify it into one of several classes (such as intent) as defined in example data provided at startup time.

The user provides example utterances in text files. There is one file per class (that is, per intention) and one line per example utterance (that is, per example user request). The directory containing the data files is defined in file utterance_classifier.py, the line starting with DATA_DIR.

• Edit the file utterance_clasifier.py (see Data format)
• Run ./utterance_classifier.py