jaotheboss / abstractive-summarisation

Using a deep learning model that takes advantage of LSTM and a custom Attention layer, we create an algorithm that is able to train on reviews and existent summaries to churn out and generate brand new summaries of its own.

PSA Data Analytics Business Challenge

This challenge was posed by PSA; Singapore Ports of Authority.

Context:

• We were given data on the usage and platform traffic of PSA's proprietary software - PortNet
• The data focused on the Haulier demographic
• The data detailed the different actions these Hauliers used on PortNet
• The data included the timing of these actions, which user performed the action, which organisation this user is from
• We were given a separate data file detailing the browser and software platform that the user used to access PortNet

Objective:

• To analyse the data to find user patterns
• To analyse the data to find the affects of Covid-19
• To find ways and means to enhance the user experience
• Many more. (Refer to the Context page of the Report)

Notes:

• Both Python and R codes were used for the analysis. R was used mostly for some visualisations.
• Bulk of the analysis and cleaning was done in Python
• This Git should consist of several Python and R files, and only 1 PDF file detailing our findings and efforts.

Link to Raw Data

abstractive-summarisation's Issues

Dataset

Where can I find the dataset?