Example projects

The Carbon Budget

The UK government has set the target for the country to achieve “Net Zero” carbon emissions by 2050 to tackle the climate crisis. To reach this goal, we may need some radical and disruptive interventions. The idea of the Carbon Budget is simple: every person has a carbon budget that they can spend each month.

In this project you will design, develop and test an App that lets people manage their carbon budget, called the Carbon Budget Wallet. The Carbon Budget Wallet (CBW) should have a user friendly interface (web and mobile) for people to enter the carbon-emitting activities undertaken and products purchased/consumed. The app will then calculate the impact on the budget and update the budget on the basis of the items entered. The app could also have smart features such as forecasting and advice, predicting when the monthly budget will be depleted, and provide advice on which products and activities have a lower carbon footprint. The app could make use of information visualisation to allow users to monitor and inspect their budget ‘transactions’ that allows them to examine the impact of their consumer choices over time.

Social media sentiment analysis

How trustworthy are autonomous systems such as robots, driverless cars, and contact-tracing apps? This project involves making use of NLP (Natural Language Processing) to analyse `sentiment’ of publicly available statements by the news media and the general public, such as from Twitter or Facebook. Such a project involves,

  • obtaining a data set that includes relevant data for your topic;
  • pre-processing of the data, such as to ‘clean’ the data;
  • develop appropriate (context/domain-specific) classification model for the data (e.g., feature selection);
  • analyse the sentiment of the data following machine learning-based techniques.

A News Aggregator and Analyser to Track Public Sentiment

In this project you will build an application that will track, download and aggregate news about a specific set of topics in a pre-specified set of British newspapers. This application will provide a dashboard that allows the user to track the subtopics being discussed as well as the overall sentiment expressed in the news articles based on time (publication date) and space (allowing the user to specify a location for each news outlet). It will be evaluated by real linguists and scientists and might end up being part of a larger project.

An application for automated time segmentation of text corpora

Analysing language change over time is a common challenge in corpus linguistics. In this project, you will build an application (either web-based or desktop) that can analyse a dataset of timestamped text documents and map trending topics based on multiple methods (bursty ngrams, topic modelling) in order to allow for the segmentation of the dataset in contiguous time segments. As part of the project you will have to investigate and define appropriate trend detection methods, build the application, and validate the application with a small case study.

Factchecker plugin

Develop a Chrome plugin that grabs the content of a webpage visited, pulls out keywords and searches amongst fact checking websites for relevant fact checked stories. The Google FactCheck Explorer tool https://toolbox.google.com/factcheck/explorer provides a tool to perform such searches across fact checking sites, and indeed they also offer an API https://toolbox.google.com/factcheck/api that the project could use. As a starting point we already have the “Pripa” prototype browser plugin that provides a framework for processing the text in a page that can be extended… (see project above). This project requires knowlege of Javascript, HTML and CSS.

Smarter charging

The Carbon Intensity service (https://www.carbonintensity.org.uk) offers an API to obtain predictions of electricity grid carbon intensity. Considering specifically at home electric vehicle charging as a use case, the project is to write an optimization system that given a predicted demand (a schedule of upcoming electric car journeys), uses the carbon intensity API to generate the best times during which to charge the vehicle to minimise the carbon footprint and continuously update it based on unanticpated usage (e.g. unscheduled quick trip to shops). The demands will be based on a series of personas (9-5 office worker car communter, nurse car commuter on shift pattern, community worker, car only evenings and weekend, etc), to exercise the optimization.