Data Science master student semester project on the automation of answering questions received by ETH IT service using an internal FAQ list. For more details please refer to reports/*
modify code/run_pipeline.py to chosen experiment
run python code/run_pipeline.py
├── README.md <- The top-level README for developers using this project.
│
├── data <- Not available on Git for privacy reasons.
│
├── PreparationWork <- Jupyter notebooks.
│ Contains all data cleaning and exploratory work.
│
├── reports <- Project report, poster and presentation PDFs.
│
├── code <- Source code for use in this project.
│ ├── run_pipeline.py <- Script to run the full pipeline.
│ │ Aknowledging bad CS practices, this is the file to modify
│ │ to run different experiments
│ │
│ ├── embedding <- Scripts to embedd clean data into matrix form
│ │ ├── preprocessing.py
│ │ └── embedding.py
│ │
│ ├── similarity <- Scripts to tag ticket answers using unsupervised ML
│ │
│ ├── classifier <- Scripts to classify ticket questions using tagged FAQ using
│ │ multiclass classification.
│ │
│ ├── visualisation <- Scripts to visualise embeddings
│ │
│ └── archive <- Archive of older attempts - to be ignored
│
└── app <- Final algorithm visualisation