Live notes, so an incomplete, partial record of what actually happened. Also, some material presented at this event was not intended to be shared, so notes may be patchy
Tags: dhbenelux
Site: http://2019.dhbenelux.org/
My asides in {}
Tweet embeds are things I liked that seemed relevant to include or that captured things I missed
Are we talking about documenting research practices because we live in an age that wants radical transparency? And a time when we care as much about process (is the banana free range) as the product (it is a banana)?
We in DH don't have a culture of valuing reproducing and validating research. And the big tent doesn't help this.
How do we get gratification out of documentation?
For whom to we document? Should we document failure? What disciplines can we learn from?
### Welcome
DH Benelux has a journal .. focused on integration
### History & Networks
Abstract .. Antonio Magliabechi, librarian, 1650-1714 .. interactions between opennees and closure in the Republic of Letters .. focus on networks and their temporality .. using card catalogues of correspondence, metadata (so not the content of the letters) .. notion of brokerage and closure in networks (Ronald S. Burt (2007)): brokers stand between communities, connect them, provide vital links, have social capital; people in closed positions (centre of a network) present as trustworthy and create protection .. over the course of her career, Magliabechi moved from closed networks (tight, local social cohesion) to becoming a broker between networks, before - towards the end of his life - introducing other Italian colleagues to take on this brokerage role.
Loving Van Vugt's insight on how 'brokers' in the Republic of Letters worked towards the end of their career, gradually handing over their important informal role to younger colleagues, nurturing the next generation of scholars #dhbenelux2019
— James Baker (@j_w_baker) September 12, 2019
@Inge_vanVugt on how introducing temporality in network analysis can provide us with a blueprint of the epistolary career of knowledge brokers in the Republic of Letters starting with model-role Antonio Magliabechi #dhbenelux2019 pic.twitter.com/SLRV1wp25b
— Paolo Rossini (@prossini1991) September 12, 2019
Movie Circulation in Four European Cities (1952): A Network Analysis of Film Programming (van Oort, Pafort-Overduin, Lotze and Jernudd)
Antwerp, Bari, Gothenburg, Rotterdam: second-tier cities .. 1952 .. looking at things like cinema programming .. some cinemas a 'givers' and some 'receivers' .. this methods doesn't allow for time, e.g. simultaneous premiers.
Network viz are projections of data not evidence .. networks of illustrations, mid to late nineteenth century .. transnational reuse of illustrations .. only looked at 1863, dresses and mantels, by name .. Der Bazaar and reprints in places like La Mode (Spanish) and La Moda (French) .. prints that look the same seem to be moving West, though if that is stereotypes or copies/piracy is tricky to disentangle ..
#dhbenelux2019 @juliebirkholz talking about impressions from stereotyped fashion plates moving between places (westward) over time. I think @willfinley91 might find this interesting.. pic.twitter.com/UYNz5VnvYx
— James Baker (@j_w_baker) September 12, 2019
We as academics don't control the interface to the datafication of our objects of study .. historians are the worst capitalist hoarders of them all .. we were trained to study through librarians and archivists, but the search box strips our objects of study of that context .. pre-mid 90s, we ignored and denied the power of the archive ~ 'our blinkered eyes were shaped by a machine for knowing (the library, the archive), the dirty little secret of humanities research' .. we are trapped in western ways of doing and thinking, amplified by Google et al .. take journal articles: students are presented with them radically decontextualised from their circumstances of production .. there is no good representation of the ecosystem of knowledge in which we operate ..TNA: home office archives much better catalogued than military archives (even though the latter are in the majority) ..
We don't know the "ecosystem of knowledge" we evolve in and we need to create ways to explore it at global and close level : we need a radical contextualisation #dhbenelux2019
— Estelle (@EstelleSzmidt) September 12, 2019
<script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>In an old school library the underlying system of knowledge was transparant, today the current systems of search and discovery effectively hide and distort the context of research data. 🤔 We are turning into servants of the beast, the machine. 😳 @TimHitchcock #dhbenelux2019
— Steven Claeyssens (@sclaeyssens) September 12, 2019
.. 3d puppet show example: https://oldbaileyvoices.org/puppetshow.php?trialid=t18530613-718 .. when we read both close and far away we satisfy the requirements of a macroscope .. until we acknowledge the western structural bias of what we have available to us, we won't be able to comprehend the non-Western .. library and archival science have been much more richly engaged with the biases of our inherited heritage .. unless we stop lying we will fail do to effective scholarship.
I love the way @TimHitchcock turns the library catalogue into an object of research, to reveal the bias we have in the Western world, the limits of what we know, to show how the world of libraries and the world of research are intimately connected. Etc. etc. #dhbenelux2019 pic.twitter.com/kkHpiI7Hob
— Steven Claeyssens (@sclaeyssens) September 12, 2019
Improving the Training of Deep Convolutional Neural Networks for Art Classification: from Transfer Learning to Multi-Task Learning (Sabatelli, Geurts)
Most deep learning trained on photorealistic images, not things like paintings .. not enough data if we only use images from an artistic domain .. transfer learning is to use similar datasets to build the model to use on an artwork .. can the CNN classify material, type, and artist, but just using the Rijksmuseum dataset didn't produce encouraging results .. so they blended standard ImageNet problem with domain specific data ..
1.5 million words of Mary Dorothy George: a computational approach to curatorial voice and legacy descriptions of art objects (Baker & Salway)
Making sense of non-sense. Tracing topics in a historical corpus on psychiatry facing low OCR quality (Biryukov, Wieneke and Andersen)
"Making sense of non-sense. Tracing topics in a historical corpus on psychiatry facing low OCR quality" project by @_evaandersen (@dtu_dhh) and Lars Wieneke and Maria Biryukov from @C2DH_LU #dhbenelux2019 pic.twitter.com/Sx2Mp4HmBS
— Katya Kamlovskaya (@kamlovskaya) September 12, 2019
Corpus work as a way into general paralysis as a way into the German influence on the field in the late-C19, which is not in the existing literature.
Cool presentation on Tracing topics in a historical corpus on psychiatry facing low OCR quality. I do appreciate further perspectives of the whole project! #dhbenelux2019 @_evaandersen @dtu_dhh @C2DH_LU pic.twitter.com/dYXasv8NQV
— Jakub Bronec (@Jakubxy) September 12, 2019
#### Mining memories of the Indonesian War of Independence (Verhaar)
Nature and extent of war crimes perpetuated by the Dutch during the 1985 Indonesian War of Independence .. c.8000 instances of war crimes in the database, extracted from ego documents .. research questions: Dutch attitudes towards Indonesians + detect acts of violence and war crimes .. gold standard of hand marking of acts of violence complimented by machine methods.
Really nice afternoon session at #dhbenelux2019 where we got to discuss the relationship between results produced using computational methods with results using more traditional methods.
— Helle Strandgaard Jensen (@Sesamescholar) September 12, 2019
Helle Strandgaard Jensen – Money, Moral and Representation: The day Stuart Hall joined my Archives 101 class
Identify as a digital historian rather than digital humanist: doing digital history gives historians less chance to escape (DH gives them a way out) .. focus on digital archival literacy, both by historicising archiving and working with archives ..
Digital historians are go-to people for other humanities scholars working with digital sources, because they know archives and source-criticism @Sesamescholar #dhbenelux2019
— DH Benelux (@DHBenelux) September 13, 2019
.. historians like to read history, so giving historians a history of digitisation cuts through better than just talking about how digitisation works ..
"improving digital archival literacy by historicising - that is, speaking to historians in a language they understand" @Sesamescholar #dhbenelux2019
— Katya Kamlovskaya (@kamlovskaya) September 13, 2019
.. we have remediated the archive into a cyberinfrastructure (Smithies, 2017), but historians not know about this infrastructure, and therefore use their sources poorly, unethically. How do we solve this problem? Tim suggested macroscope, Helle suggests turning to Stuart Hall .. meaning making in media .. see archive as a medium that encodes certain says of knowing .. historians do a lots of thinking about the logics of archives, because going to the archive is the heartbeat of historical research. Which makes it all the more curious that they don't engage with the logical of the digital archive .. cultural institutions make it far too easy to find things, and in so doing remove their labour from the process. we need to make finding things more difficult! .. Four Big Problems ..
The importance of demystifying digitisation and infrastructure says @Sesamescholar Absolutely! #dhbenelux2019 pic.twitter.com/PVCcIzTuvX
— Sally Chambers (@schambers3) September 13, 2019
.. 1) digitisation projects are driven by strategic needs or external funding. So we tend to recanonise things that don't need recanonising. Historians need to be more aware of selection processes .. 2) search is easy and makes methodological implications invisible .. 3) metadata standardises in a way that creates simplicity. Georgian Papers Programme excellent example of refusing to do this, so that first-peoples can be found .. 4) virtually no explanation of the process of creating digital archives, because the focus on outreach limits the scope for documentation ..
At #dhbenelux2019 @Sesamescholar gives an interesting example of competing efforts between the outreach interfaces and more traditional institional interfaces of archival collections : User friendly vs. Transparent about provenance and content overview pic.twitter.com/QEQRsJ7Cil
— Estelle (@EstelleSzmidt) September 13, 2019
.. we need to work more closely with GLAMs to support them in making documentation. We need to teach archival literary.
"We as a community should help institutions be better at documenting projects" and facilitate "transparency over easy access". @Sesamescholar #dhbenelux2019 pic.twitter.com/2ytZcorr3a
— Katya Kamlovskaya (@kamlovskaya) September 13, 2019
It's all about documentation. @sesamescholar: Institutions have to make the process of digitisation and publication more transparant. What is being digitised? What's behind the search field? What metadata is there? How was the object transformed? #dhbenelux2019 Agreed! But... 1/2
— Steven Claeyssens (@sclaeyssens) September 13, 2019
Supporting the Interpretation of Enriched Audiovisual Sources through Temporal Content Exploration (Huurdeman, Melgar Estrada, Noordegraaf)
Media Suite: Dutch multi-media research environment .. automated speech recognition (ASR), speech to text .. presenting ASR transcripts alongside AV .. temporary tag clouds attached to parts of a video to aid browsing: user testing underlined there was both positive and negative serendipity at play in their use .. annotating video as a timeline ..
The paradox of “textualizing” digitized audiovisual sources: Automatic speech recognition in digital television collections (Tom Slootweg)
Abstract .. working on computer vision and Automated Speech Recognition (ASR), and the paradoxes of using those techniques in the context of media studies .. perhaps we need an auditive digital turn, break the dominance of the visual .. but is ASR truly auditive? It is textualising and reducing complexity .. how much of the data file is actually being used to create ASR? What is left over? audio fingering printing going on at Technical Institute Zurich ..
Dominant medium for news and entertainment .. metadata as cultural objects in order to understand where radio collections have ended up, track them through time and space ... 1930s growth in purpose built radio recording studios that had archives within them, and processes for collecting and organising .. and these systems of archiving have longevity .. after WW2, there was plunder and confiscation of radio collections, but if you go to most radio archives in Europe you'd never have a sense that the materials had left the building, provenance is just not there ..
Ha ha, since the vibe of the conference #dhbenelux2019 was so good, and working with my co-chair @ClaartjeR was so much fun, recovering didn't take long. https://t.co/enMa2rj58h
— susan aasman (@aasmanna) September 13, 2019
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Exceptions: embeds to and from external sources, and direct quotations from speakers