Main Page
Welcome to the wiki of the course Foundation of Digital Humanities (DH-405).
Contact
Professor: Frédéric Kaplan
Assistant: Vincent Buntinx
Rooms: Wednesday (CM1113) and Friday (CM1104 and DIA005)
Links
Summary
This course gives an introduction to the fundamental concepts and methods of the Digital Humanities, both from a theoretical and applied point of view. The course introduces the Digital Humanities circle of processing and interpretation, from data acquisition to new understandings and services. The first part of the course presents the technical pipelines for digitising, analysing and modelling written documents (printed and handwritten), maps, photographs and 3d objects and environments. The second part of the course details the principles of the most important algorithms in particular deep learning approaches (for document analysis and image generation) and knowledge modelling (semantic web, ontologies, graph databases). The third part of the course focuses on platform management from the points of view of data, users and bots. Students will practise the skills they learn by engaging in a class-wide collective project.
Plan
Introduction
Week 1 : Structural tensions in Digital Humanities
19.09 (2h) Introduction to the course and Digital Humanities, structure of the course. Introduction to Framapad with a simple exercise. Principle of collective note talking and use in the course. State of the Digital Humanities at EPFL, in Switzerland and in Europe. Structuring tensions 1: Digital Humanities, Digital Studies, Humanities Computing and Studies about Digital Culture. Digital Humanism vs. Digital Humanities. Why digital methods tend to dissolve traditional disciplinary frontiers. A focus on practice. Translation issues.
21.09 (2h) (a) Structuring tensions 2: Big Data Digital Humanities vs Small Data Digital Humanities. The 3 circles. Exercise on relationship between elements in Digital Culture schema. (2h) Practical session: Introduction to MediaWiki. Objective: Learning the basic syntax of MediaWiki. Get a first experience of collaborative editing. Learning to write from a neutral point of view. Creation of a Wikipedia page on the concepts of Linguistic Capitalism, Big data of the past, Crowdsifting and Thick mapping
Week 2 : Patrimonial capitalism and common goods
26.09 (1h) Introduction to the DH circle linking the digitisation of sources, their processing, their analysis, visualisation and the creation of societal value (insight, culture) leading ultimately to the digitisation of new sources. Presentation of some sustainable DH circles (genealogy, image banks). Patrimonial capitalism and the risk of monopolistic companies. Parallelism with the race for sequencing the Human Genome. Introduction to the TIme Machine FET Flagship and mutualised infrastructure approach. (1h) General presentation of the Time Machine pipeline at the Datasquare / ArtLab pavillon.
28.09 (4h) Paris : Bibliothèque nationale de France.
Part I : Pipelines
Week 3: Documents pipeline
03.10 (2h) The Digitization Process and Pipelines. What is a document? What is a digital image? Exercise on Book Scanners typologies. Document digitisation as a problem of conversion of dimensions. Digitisation is logistic optimization. Alienation. Digitisation on demand. Fedorov's notion of optimal experiment.
05.10 (2h) Pipeline for Written documents. Part I: Standards. Open Annotation Data Model. Shared Canvas. Part 2: Regulated Representations. DHCanvas. (2h) Presentation of the Projects. Research of 5 maps of maps series. Formation of the groups
Week 4: Artworks Pipeline
10.10 (2h) Pipeline for Artworks photographs (Benoit Seguin). Image banks and phototarchives. Scanning techniques for photographs. Segmentation. Visual similarity vs visual connections.
12.10 (4h) Introduction to deep learning approaches (Benoit Seguin). Exercises with the Replica database and search engine. 5 mn presentation. Work on projects. Formation of the groups.
Week 5: Maps Pipeline
17.10 (2h) What are cartographic documents. Exercice on ancient maps. History of cartography. Odometry. Triangulation. Coordinate systems. Metric systems. Projection. Cadaster. Aerial photography.
19.10 (4h) (a) Introduction to GIS. Points, Lines, Polygons. Coordinate Sytems. QGIS Hands On. Exercise on Venetian cadaster (Bastien). (b) Oral presentation of project plan (ungraded). All project validated.
Week 6: 3D Pipeline
24.10 (2h) Pipeline for 3D spaces. Modelling vs Sampling : Part I : Modelling. Photogrammety. Demo Sketchup. Model-based Procedural methods. Architectural grammars. Class I and Class II elements. The question of realism.
26.10 (4h) Part II : Sampling. Photogrammetric tutorial (Nils)
Part II : Algorithms
Week 7 : Deep Learning algorithms
31.10 (2h) Time Machine 2018 http://conference.timemachine.eu
02.11 (2h) A panorama of Deep learning methods. Successes. Fundamental principles. Neurons. Receptive Fields. Hierarchical representation / texture. Gradient descent. Credit Assignment Path. Most important architectures. Convolutional neural networks. Recurrent neural networks. Siamese Networks. Word2Vec. Generative Adversarial Networks. Style Transfer. Importance of Deep learning for Digital Humanities. Can Deep Learning networks and Big Data of the Past lead to new forms of Artificial Intelligence ? Machine vision tutorial (Benoit, Sofia). Introduction to Anaconda, Jupyter, TensorFlow. Deep learning in practice.
Week 8 : Knowledge modelling
07.11 (2h) The beauty of Knowledge modelling. Tables. Databases. Semantic web, Ontologies, URI, RDF, CIDOC-CRM, How to code event, places and influence. Metaknowledge
09.11 (4h) (a) Exercise in semantic modelling and inference (Maud). Graph writing. Presentation of some interesting ontologies: SKOS, VIAF, Geonames, TGN, W3C Time Ontology. SPARQL and SPARQL endpoint. Exercice on SPARQL endpoints: DBPedia [1], Talk of Europe [2], Persée [3], Le Temps ARchive [4]. (b) Work on project, preparation of presentation
-- Project plan and milestones deliverable on the Wikipage of each project (10%)
Week 9 : Project
14.11 (2h) Midterm presentation (10%)
16.11 (4h) Project development
Part III : Platform management
Week 10 : Data Management
21.11 (2h) Data Management : FAIR principle, Creative Commons, Data Management models, Sustainability, Right to Forgotten. Management of uncertainty, incoherence and errors. Iconographic principle of precaution.
23.11 (4h) ----
Week 11 : User Management
28.11 (2h) Programming and Testing phase : Production of first Pulses
30.11 (4h) User Management : Part I: Persona. Part II: Motivation and onboarding dynamics. Three case studies: Twitter. Quora. Wikipedia. Part III: "Wisdom" of the crowds. Collectivism vs Liberalism. Open source as a form of liberalism for engineering. The ambiguous of fork. Part IV: The "power" of the crowds. Mechanical Turk. Crowdflower. Crowdfunding
Week 12 : Bot Management
05.12 (2h) Bot Management : Three case studies on bot management : Twitter, Wikipedia, Google.
07.12 (4h) Testing phase and report writing
12.12 (2h) Report writing
14.12 (4h) Discussion of the collective paper
-- Deadline for GitHub repository (10%)
19.12 (2h) Final project presentation (20%)
21.12 (2h) ----
References
Assessment and Notation grid
- 2 oral presentations (30%)
- 1 midterm presentation of the project (10%)
- 1 final discussing the project result (20%)
- Written deliverables (Wiki writing) (40%)
- Quality of the project (30%)
2 collective oral presentations (30%)
Midterm presenting the project planning and prior art 10' (10%)
Notation grid :
- The presentation contains a planning and discussion of prior art (4)
- + 0.5 The slides are clear and well presented
- + 0.5 The oral presentation is dynamic and fluid
- + 0.5 The planning is realistic.
- + 0.5 The students answer well to the questions
Final discussing the project result 20' (20%)
Notation grid :
- The presentation presents the results of the project (4)
- + 0.5 The slides are clear and well presented
- + 0.5 The oral presentation is dynamic and fluid
- + 0.5 The results are will discussed
- + 0.5 The students answer well to the questions
Written deliverables (Wiki writing) (40%)
- Projet plan and milestones (10%) (>300 words)
- Historical introduction to the map (10%) (>200 words)
- Detailed description of the extraction methods (10%) (>500 words)
- Quantitive analysis of the performances of extraction (10%) (>300 words)
- Motivation and description of the services (10%) (>200 words)
Production (30%)
- Quality of the realisation 20%
- Code deliverable on github 10%