Before turning to how the staff members accomplish this task, it is worth illustrating how conversational topics are generated in the TC meetings more generally.
When running on different datasets, LDA suffers from "order effects", i.e., different topics are generated if the order of training data is shuffled.
In order to form an agenda for future research in the area of biofouling and the marine renewable energy industry, 119 topics were generated, categorised and prioritised.
For each of these areas several topics were generated.
From the transcriptions, 40 potential topics were generated using the Mayring method of qualitative content analysis [ 25].
Topics are generated via a wide range of stakeholder groups and sources, such as web submissions, workshops with clinicians, academics and patient and public representatives and outputs from the James Lind Alliance Priority Setting Partnerships (http://www.lindalliance.org/).org/
A range of themes relating to this topic were generated in the analysis.
The piloted topic guide was generated through close consultation with local sexual health and HIV prevention services, including non-governmental organisations, and fostered participant led description of experience related to sex work.
A pool of statements that could potentially describe and sufficiently represent the topic of investigation was generated.
The above equation states that, for each word in the query, we compute the overall probability that it was generated by any topic, given the topic proportions in the potentially relevant experiment.
