Motivation

Research in the field of Digital Humanities, also known as Humanities Computing, has seen a steady increase over the past years. Situated at the intersection of computing science and the humanities, present efforts focus on building resources such as corpora of texts, images, musical pieces and other semiotic artifacts digitally available, searchable and analyzable. To this end, computational tools enabling textual search, visual analytics, data mining, statistics and natural language processing are harnessed to support the humanities researcher. The processing of large data sets with appropriate software opens up novel and fruitful approaches to questions in the 'traditional' humanities. Thus, the computational paradigm has the potential to transform them. One reason is that this kind of processing opens the way to new research questions in the humanities and especially for different methodologies for answering them. Further, it allows for analyzing much larger amounts of data in a quantitative and automated fashion - amounts of data that have never been analyzed before in the respective field of research. The question whether such steps ahead in terms of quantification lead also to steps ahead in terms of the quality of research has been at the core of the motivation of the seminar.

Obviously, despite the considerable increase in digital humanities research, a perceived gap between the traditional humanities and computer science still persists. Reasons for this gap are rooted in the current state of both fields: since computer science excels at automating repetitive tasks regarding rather low levels of content processing, it can be difficult for computer scientists to fully appreciate the concerns and research goals of their colleagues in the humanities. For humanities scholars, in turn, it is often hard to imagine what computer technology can and cannot provide, how to interpret automatically generated results, and how to judge the advantages of (even imperfect) automatic processing over manual analyses.

To close this gap, the organizers proposed to boost the rapidly emerging interdisciplinary field of Computational Humanities (CH). To this end, they organized a same-named Dagstuhl Seminar that brought together leading researchers in the fields of Digital Humanities and related disciplines. The seminar aimed at solidifying CH as an independent field of research and also at identifying the most promising directions for creating a common understanding of goals and methodologies.

At the core of the organizers' understanding of CH is the idea that CH is a discipline that should provide an algorithmic foundation as a bridge between computer science and the humanities. As a new discipline, CH is explicitly concerned with research questions from the humanities that can more successfully be solved by means of computing. CH is also concerned with pertinent research questions from computing science focusing on multimedia content, uncertainties of digitisation, language use across long time spans and visual presentation of content and form.

In order to meet this transdisciplinary conception of CH, it is necessary to rethink the roles of both computer scientist and humanities scholars. In line with such a rethinking, computer scientists cannot be reduced to software engineers whose task is just to support humanities scholars. On the other hand, humanities scholars cannot be compelled to construe post-hoc explanations for results from automatic data analysis. Rather, a common vision - shared among both groups of scientists - is needed that defines and exemplifies accepted methodologies and measures for assessing the validity of research hypotheses in CH. This vision motivated and formed a common ground for all discussions throughout the seminar.

Goals and Content of the Seminar

In order to elaborate the vision of CH as a bridge between computer science and the humanities, the seminar focused on questions that can be subsumed under four different reference points of problematizing CH:

1. The Present State: What works, what does not?
   • Review of the success of the last 10 years of the digital humanities: Can we identify commonalities of successful projects? What kinds of results have been obtained? What kinds of results were particularly beneficial for partners in different areas of research? Can success in one field be transferred to other fields by following the same methodology?
   • Review of the challenges of the last 10 years of the digital humanities: What are recurring barriers to efficient cross-disciplinary collaboration? What are the most common unexpected causes of delays in projects? What are common misunderstandings?
   • What is the current role of computer scientists and researchers in the humanities in common projects, and how do these groups envision and define their roles in this interplay?
2. Computational Challenges in Computational Humanities:
   • What research questions arise for computational scientists when processing data from the humanities?
   • How can the success of a computer system for humanities data-processing be evaluated to quantify its success?
   • What are the challenges posed by the demands from the humanities? In particular, how can computer scientists convey the notion of uncertainties and processing errors to researchers in the humanities?
3. Humanities Challenges in Computational Humanities:
   • What research questions can be appropriately addressed with computational means?
   • How can we falsify hypotheses with data processing support?
   • What is and is not acceptable methodology when one relies on automatic data processing steps?
4. Common Vision: Algorithmic Foundations of Computational Humanities:
   • Can we agree on generic statements about the expressivity of the range of algorithms that are operative in the digital humanities and related fields of research?
   • Can we distinguish complexity levels of algorithms in the computational humanities that are distinguished by their conditions of application, by their expressiveness or even explanatory power?
   • Which conditions influence the interpretability of the output generated by these algorithms from the point of view of researchers in the humanities?

The Program

In order to work through our set of goals (see Section 1), the seminar decided for a mixture of talks, working groups and plenary discussions. To this end, four Working Groups (WG) have been established whose results are reported in respective sections of this report:

• The Working Group on Ethics and Big Data (members: Bettina Berendt, Chris Biemann, Marco Büchler, Geoffrey Rockwell, Joachim Scharloth, Claire Warwick) discussed a very prominent topic with direct relationships to recent debates about ethical and privacy issues on the one hand and the hype about big data as raised by computer science on the other. One emphasis of the WG was on teaching how to process big data, how this research relates to legal and ethical issues, and how to keep on public dialogs in which such issues can be openly discussed -- beyond the narrow focus of the academic community. A central orientation of this discussion was to prevent any delegation of such discussions to closed rounds of experts ('research ethics boards') which do not support open discussions to a degree seen to be indispensable by the WG. The widespread, fruitful and detail-rich discussion of the WG is reported in more detail in Section 4.1.
• The Working Group on Interdisciplinary Collaborations -- How can computer scientists and humanists collaborate?
  (members: Jana Diesner, Christiane Fellbaum, Anette Frank, Gerhard Heyer, Cathleen Kantner, Jonas Kuhn, Andrea Rapp, Szymon Rusinkiewicz, Susan Schreibman, Caroline Sporleder) dealt with opportunities and pitfalls of cooperations among computer scientists and humanities scholars.The WG elaborated a confusion matrix that contrasts commonplaces and challenges from the point of view of both (families of) disciplines. Ideally, scientists meet at the intersection which challenges both groups of scientists - thereby establishing CH potentially as a new discipline. In any event, this analysis also rules out approaches that reduce either side of this cooperation to the provision of services, whether in terms of computing services or in terms of data provisions. More information about the interesting results of this working group are found in Section 4.2.
• The Working Group Beyond Text (members: Siegfried Handschuh, Kai-Uwe Kühnberger, Andy Lücking, Maximilian Schich, Ute Schmid, Wolfgang Stille, Manfred Thaller) shed light on approaches that go beyond language in that they primarily deal with non-linguistic information objects as exemplified by artworks or even by everyday gestures. A guiding question of this WG concerned the existence of content-related features of such information objects that can be explored by computational methods. As a matter of fact, corpus building by example of such artifacts is in many cases still out of reach so that computation can hardly access these objects. Seemingly, any success in 'computerizing' research methodologies here hinges largely upon human interpretation. Obviously, this is a predestined field of application of human computation with the power of integrating still rather separated disciplines (e.g., musicology, history of art, linguistics etc.). See Section 4.3 for more information about this promising development.
• The Working Group on Literature, Lexicon, Diachrony (members: Loretta Auvil, David Bamman, Christopher Brown, Gregory Crane, Kurt Gärtner, Fotis Jannidis, Brian Joseph, Alexander Mehler, David Mimno, David Smith) dealt with the role of information as stored in large-scale lexicons for any process of automatic text processing with a special focus on historical texts. To this end, the WG started from the role of lexica in preprocessing, the indispensability of accounting for time-related variation in modeling lexical knowledge, the necessity to also include syntactic information, and the field of application of automatic text analysis. Special emphasis was on error detection, correction and propagation. The WG has been concerned, for example, with estimating the impact of lemmatization errors on subsequent procedures such as topic modeling. In support of computational historical linguistics, the WG made several proposals on how to extend lexica (by morphological and syntactical knowledge) and how to link these resources with procedures of automatic text processing. See Section 4.4 for more information about the results of this WG.

Part and parcel of the work of these WGs were the plenary sessions in which they had to present their intermediary results in order to start and foster discussions. To this end, the whole seminar came together - enabling inter-group discussions and possibly motivating the change of group membership. Beyond the working groups, the work of the seminar relied on several plenary talks which partly resulted in separate position papers as published in this report:

• In his talk on Digital and computational humanities, Gerhard Heyer shed light on the role of computer science in text analysis thereby stressing the notion of exploring knowledge or text mining. He further showed how these methods give access to completely new research questions in order to distinguish between (more resource-related) Digital Humanities and (algorithmic) Computational Humanities.
• In his talk, Chris Biemann tackled the field of Machine Learning methods from the point of view of their application to humanities data. He clarified the boundedness of these methods in terms of what is called understanding in the humanities. From this point of view, he pleaded for a kind of methodological awareness that allows for applying these methods by clearly reflecting their limitations.
• In their talk on On Covering the Gap between Computation and Humanities, Alexander Mehler & Andy Lücking distinguished differences that put apart both disciplines. This includes a methodological, a semiotic and an epistemic gap that together result via an interpretation gap into a data gap. In order overcome these differences, they pleaded for developing what they call hermeneutic technologies.
• In her talk on Digital Humanities & Digital Scholarly Editions, Susan Schreibman gave an overview of her work on multimodal, multicodal digital editions that integrate historical, biographical and geographical data. Her talk gave an example of how to pave the way for a people's history in the digital age. To this end, she integrates recent achievements in data mining (most notably network analysis, geospatial modeling, topic modeling and sentiment analysis).
• In his talk on How can Computer Science and Musicology benefit from each other? , Meinhard Müller switched the topic of mainly textual artifacts to musical pieces and, thus, to musical artworks. He explained the current possibilities of automatic analysis of musical pieces and demonstrated this by a range of well-known examples of classical music.

This work nicely shows that computational humanities has the goal of covering all kinds of data as currently analyzed and interpreted in the humanities (see also the Working Group Beyond Text for such a view).

The seminar additionally included a range of short talks in which participants presented state-of-the-art results of their research: among others, this included talks by Christopher Brown, Anette Frank, Brian Joseph and Szymon Rusinkiewicz. This work nicely provided information about a range of linguistic and multimodal application areas and, therefore, reflected the rich nature and heterogeneity of research objects in the humanities.

A highlight of the seminar was a plenary discussion introduced by two talks given by Gregory Crane and by Manfred Thaller. These talks started and motivated an academic verbal dispute in which, finally, the whole seminar participated in order to outline future challenges of Digital Humanities with impact beyond the border of these disciplines -- even onto the society as a whole. Both talks - on Evolving Computation, New Research Directions and Citizen Science for Ancient Greek and the Humanities by Gregory Crane (see Section 5.1) and on The Humanities are about research, first and foremost; their interaction with Computer Science should be too by Manfred Thaller (see Section 5.2) - opened a broad discussion about the role of humanities among the sciences and their status within the society.

Last, but not least, we should mention two common sessions with a concurrent seminar on Paleography. These sessions, which took place at the beginning and at the end of the seminars, opened an interesting perspective on one particular field that could be counted as a sub-discipline of Computational Humanities. The paleographers met in Dagstuhl for the second time and discussed some of our CH issues previously; it was fruitful to exchange approaches on how to overcome them.

Conclusion

Most of the working groups used their cooperation as a starting point for preparing full papers in which the theme of the group is handled more thoroughly. To this end, the plenary discussed several publication projects including special issues of well-known journals in the field of digital humanities. A further topic concerned follow-up Dagstuhl seminars. The ongoing discussions around the perceived gap between computer science and the humanities and the various proposals from the participants on how to define, bridge or deny this gap made it clear that the seminar addressed a topic that needed discussion and still needs discussion. The talks, panels and working group discussions greatly helped in creating a better mutual understanding and rectifying mutual expectations.

In a nutshell: the participants agreed upon the need to continue the discussion since CH is a young and open discipline.

Creative Commons BY 3.0 Unported license

Chris Biemann, Gregory R. Crane, Christiane D. Fellbaum, and Alexander Mehler