An Introduction to Computational Text Mining and Topic Modelling

RSVP HERE

The abundance of textual data available nowadays from social media, online databases, digital archives and other sources present researchers with exciting opportunities as well as formidable challenges. There is often more text available than any researcher can hope (or would want) to read. Thankfully, there are computational tools available to assist the navigation and analysis of such textual datasets. These tools can do much more than count keywords. For example, named entity recognition algorithms can tell the difference between the names of people, places and organisations. Another class of algorithms, known as topic models, can identify thematic categories in a dataset based solely on how words co-occur.

Such analytical power, however, does not come for free. These tools can be tricky to use yet easy to abuse. Their successful application demands not only technical competence but also critical vigilance to guard against spurious interpretations or biased conclusions.

This seminar will provide an overview of selected computational text analysis methods, focussing on named entity recognition and topic modelling. As well as showing examples of what these methods can do, the seminar will discuss some of the issues that can arise from their use and abuse. The seminar will also review some of the software options through which researchers can access these tools.

Dr Martin Schweinberger, School of Languages & Cultures, UQ

This talk elaborates on the methods discussed by Dr. Veitch by showing how advanced computational text analytics are applied to language data using selected analyses from his research.

Format: Dr Veitch will speak for 45 minutes, Dr Schweinberger for 15 minutes, discussion 30 minutes

About the Presenters

Dr Angus Veitch is currently a postdoctoral researcher in the School of Management at RMIT University, where he is researching the public acceptance of hydrogen and other low-carbon fuels. He completed his PhD at UQ on the contributions of topic modelling to social science research using coal seam gas as a case study. Before commencing his PhD, he worked for several years in the Queensland public service in natural resources policy.

Dr Martin Schweinberger is postdoctoral research fellow in language technologies at the School of Languages and Cultures. He has specialized in computational approaches to analysing language data and was affiliated with the Language Technology Group at Universität Hamburg

RSVP HERE