(HuNI): Working with the viscosity of humanities research. By Deb Verhoeven (Deakin University, Australia) The Humanities Networked Infrastructure (HuNI) initiative is a national Virtual Laboratory that is specifically designed to support humanities research practices. Rather than aim for neutrality, HuNI deliberately shows how organised information systems can encompass a humanities disposition for diversity, co-existence, complexity, interpretation and contestability.
Technology-induced bias: if we can’t avoid it, we need to measure it. By Jacco van Ossenbruggen (CWI) Tool criticism in a world of technophilia and technophobia.
Topic modelling only works if you have the right document collection! By Iris Hendrickx (Radboud University) LDA, an unsupervised topic modeling technique, is developed to detect topics in a large document collection. This technique has certain restrictions and does not work for every arbitrary document set and can fail to produce interesting topics. Why does this happen? What are the limits of LDA?
Deconstructing interface technology; media archaeology as cracking code. By Sonja de Leeuw (Utrecht University) Technology is never neutral, rather a construct, an interpretation of user expectations and navigations. My contributions discusses whether media archaeology could help researchers to discover the provenance of technology and assess the results generated by technology as defining factor.
Automatic transcription is not neutral (ADVANT). By Wyke Stommel (RU), Tom Koole (RUG), Tessa van Charldorp (UU), Sandra van Dulmen (NIVEL) en Antal van den Bosch (RU) Conversation Analysis (CA) is an approach for analyzing social conduct (how people do things) in talk-in-interaction. In CA, the researcher starts off analyzing by making detailed and elaborate transcripts of audio/video data that do not only include verbal utterances, but also things like pauses, overlap, intonation, laughter, sighing, etc. Several decades of research have shown that these aspects are meaningful phenomena in interaction. As a result, transcription is a time-consuming effort. The question is if technology – automatic speech recognition – can facilitate transcription and to what extent it is neutral. Allegedly, the technology accelerates the process, is more objective (e.g., in measuring silences) and makes it possible to work with larger corpora. This type of technology will be optionally included in ADVANT, an Advanced Video Analysis Tool we intend to develop. In this presentation we will discuss some of the implications of using technology for transcribing. We will argue that automatic transcription involves at least three non-neutral aspects: 1) it is theory-bound (like any transcript), 2) it shows a restricted set of aspects of interaction and 3) it steers research questions and agendas.