4 day introductory session presented by Dr Travis Coan and Dr Chico Camargo

 (An optional 1 day Intro to Python session will take place on Monday 17th April for those unfamiliar with the language)

Differently from traditional software, artificially intelligent software can improve performance upon ingesting increasing quantities of data. This module will introduce you to the core concepts that are needed to understand the field of Machine Learning. You will engage with the theory and gain practical experience through a series of practical workshops. In this module we will emphasize the notion and importance of data and you will learn how machines can deal with different types of data sources, ranging from text to images to networks, and all sorts of metadata.

This course will also provide a more in-depth introduction to the use of natural language processing in computational communication research. You will learn how to apply various supervised, semi-supervised, and unsupervised machine learning methods for analysing textual data. You will also be introduced to the topics of language modelling, semantic similarity, and recent advances in transfer learning. Through a series of lectures and practical applications, this section of the course aims to provide you with the tools to use “text as data” in your own research projects.

Presented by Dr Diogo Pacheco (University of Exeter)

Networks are everywhere. They connect physical things such as neurons in your brain, and computers on the internet; and abstract relationships such as friendships or even the food web. Network science provides us with a series of tools to understand the role of nodes, the implications of the network structure, and conjecture about emergent behaviours. In this module, you will learn to identify and represent networks of our interest. We will have practical sessions and will play around with Twitter data.

Presented by Dr Constantine Boussalis

As we transcend the the digital revolution, digital images are being generated and shared at an astounding pace. To give a sense of the scale, it is estimated that over 14 billion images are shared daily on social media platforms, and over 136 billion images have been indexed on Google Image Search to date.

This workshop provides an introduction to the application of state-of-the-art computational methods to analyse the content of digital images at large scales. In particular, the workshop will offer a theoretical review of important algorithms, such as convolutional neural networks,  that have proven to be remarkably effective in a wide range of applied social science research objectives involving the use of images as data. Beyond theory, the workshop will also provide practical instruction on how to apply machine learning methods in Python to perform a number of common image-as-data tasks, such as automatically detecting faces in an image, recognizing the identity of a face, as well as automatically assigning images into coherent clusters.

The workshop is particularly designed for early stage researchers and postgraduate students who are interested in learning more about how to use computers to analyse large collections of digital images.

Presented by Dr Debora Nozza (Bocconi University)

Social media platforms have become a rich source of information for social scientists, offering new and unique insights into human behavior and social dynamics. However, analyzing this vast amount of textual data can be challenging. This workshop will provide a comprehensive overview of advanced methods of text analysis and how they can be used to uncover key insights into human behavior on social media. With hands-on exercises, participants will gain practical experience in using these techniques to inform their own research or work projects. The workshop will also address the challenges of collecting and analyzing social media data, including dealing with low-resource settings. 

Presented by Dr Nicolas Gold

The contemporary research landscape is one in which online data plays a significant role in many disciplines.  Whilst often appearing to be a straightforward and easily accessible source of research data, there are a range of ethics issues to address including privacy, rights, expectations, information, and consent.  These are not necessarily straightforward to deal with in the online context.  In this talk I will discuss some of the issues involved, present a framework that may be helpful to researchers addressing these issues in their work, and review some example situations.

Presented by Dr Travis Coan and Dr Michelle Spruce

sponsored by ACCESS

Free Session: There is no cost to partake in this session

Social sensing is an approach developed in recent years to analyse unsolicited social media data to detect real-world events of interest. In social sensing, each individual in a social network acts as a sensor; their posts providing pieces of sensor data which can be used to better understand what is happening to or near that individual at a given place and time. Filtering and grouping this information by topic, time or location provides a better understanding of an event through the eyes of a social network. 

In this session there will be an introduction to how the application of social sensing has been used to successfully detect the social impacts of natural hazard events using Twitter data. There will then be an opportunity to put social sensing into practice by undertaking some analysis of Twitter data for a specific event using Python. This will include instruction on how to apply some basic text analysis and machine learning methods.

This session would be suitable for early-career researchers and postgraduate students who are interested in learning more about how to undertake analysis of social media data.

Presented by Dr. Johannes Gruber

Free Session: There is no cost to partake in this session

This workshop provides an introduction to the open-source AmCAT software suite for content analysis, with a focus on collaborative annotation projects. The first step of most supervised machine learning projects is to add annotation to training and test data. This involves defining a coding task (codebook), selecting (a sample of) the documents to be coded, distributing the documents among coders and collecting and validating the results. With the AnnoTinder module, the AmCAT software suite offers an easy-to-use solution for manual annotation projects that looks and works nicely on desktop browsers and mobile devices. In this course, you will learn how to set up AnnoTinder and use it to plan and conduct your own research.