Contenu du sommaire : Le web : source et archive

Over the last few decades, the web has established itself as a privileged means of disseminating and sharing content and experiences. This raises the question of its memory and, consequently, its archiving. Such archiving is generally based on the principle of the collection, generally automatic and periodic, of a set of IP addresses by robots. The web pages are then recorded and stored, simulating the browsing experience of an Internet user. The result is a set of pages that could, in principle, be consulted as if we were going back to the Web of the past: in reality, as a collection is never exhaustive or permanent, the archive obtained is a temporal and spatial sampling of the Web over a given period. Under these conditions, what kind of memory regime is inaugurated by archiving the Web, and what epistemology can we deduce from this to characterise the consultations we can make of it and the lessons we can learn from it? The aim of this article is to address these questions. The challenge will be to determine what the ‘noeme' of a Web archive might be, i.e. the concept or essence that configures and predetermines our understanding of this archive.

The intention to preserve and document websites and content accessible on the Web in order to enable the writing of history, began shortly after the Web's emergence. While the overall objective of Web archiving—with its inherent challenges—is clear, defining what exactly should be preserved remains complex. Within the French legislative framework, this aim aligns with the obligation of legal deposit, specifically the legal deposit of digital documents. In the United States, the Internet Archive presents itself as a digital library of websites and other cultural artifacts in digital format. These examples illustrate the diversity of approaches, all of which establish systems dedicated to publications rather than traditional archives. The materials collected in various Web archiving projects do not fit into any standard definition of the concept of archive. Moreover, computer specialists and the information technology field in general is multiplying the use of this concept in widely varying interpretations. This paper investigates the meaning of the archive concept as applied to digital materials since the emergence of the Web. From digital traceability to computational archiving, including “open archives” initiated with the Web, the semantic scope of the archive has continually expanded alongside technological advancements. This expansion also encompasses the tools, objects, and actors that shape knowledge infrastructures through which our understanding of the digital past is constructed. The archival value of digital materials integrated into these infrastructures remains an already prefigured horizon that is yet to be fully conquered.

Based on the collaborative project Le goût de l'archive à l'ère numérique, this article notes that Arlette Farge's book Le goût de l'archive, which is very popular with French historians, no longer corresponds to the research experience of a large proportion of them. Web archives provide an interesting example of this discrepancy: often unknown to historians who could make use of them, and sometimes methodologically abused, the uses, misuses and non-uses of web archives show the extent to which a common digital culture among historians is lacking, reminding us of the need to rethink our training. The case of web archives also shows that we need new philologies, adapted to the digital age and, more specifically, to natively digital sources. As the first type of natively digital source to be widely accessible, web archives herald the challenges that contemporary historians will have to take up: firstly, to make explicit discrete digital practices which otherwise risk introducing many unanswered questions into our methodology. This clarification could go so far as to reconsider what we take for granted today, the definition of an ‘archive'.

The article delves into the utilization of the web as a source for contemporary historical research. Despite the widespread practice of consulting digitized archives, a limited number of historians employ the web as a primary source. The discussion addresses the complexities inherent in using born-digital documents and underscores the imperative of leveraging web archives for digital historiography. It highlights that these archives, reconstructed from the web, demand heightened scrutiny and a redefined relationship with archival materials. Furthermore, the article illustrates how the integration of web archives since the early 2000s has revitalized historical methodologies in conjunction with digital humanities, drawing on studies of memorial practices and collective mobilizations.

This article arises from a reflection initiated during the cycle of workshops of the ResPa
Don project, which continued through a panel discussion organized as part of the conference “The web, source and archives” in April 2023. Going back to the methodological foundations laid by the Introduction to Historical Studies by Langlois and Seignobos, it comments the notion of “auxiliary science”, questioned in the light of the scientific analysis of web sources. In a context where historians of the present time are confronted with the challenges of digital culture, we feel a growing need to create links with the professionals responsible for archiving these sources of a new type. Three paths converge around this reflection: those of Emmanuelle Bermès, assistant professor at the École des chartes and specialist in the heritage of digital collections; of Sophie Gebeil, researcher at the TELEMMe laboratory and historian of online memorial practices; and of Grégory Miura, general curator of libraries and president of ISO technical committee 46. These complementary perspectives offer an in-depth analysis of the heritage of digital collections and the study of web archives as sources. The use of auxiliary sciences as a reading grid highlights the need to re-evaluate methods of citation and stabilization of web sources, as well as the need for training in digital literacy for all researchers, with a view to development. towards standardized practices.

The article proposes to study the epistemological and technical issues of a scientific project hosted in residence at the National Library of France DataLab and entitled, Millenium : the Year 2000 problem (Y2K). Expression of imagination and technical knowledge in a pre-digital society. Based on a multimodal corpus, the field of inquiry illustrates the augmentation of web archives corpus by audiovisual via the practice of screencast and adress the implications for the study of a sociotechnical controversy.

The purpose of this study is to understand the plurality of attempts to build up a corpora of terrorist propaganda. It displays the different methodologies observed and practised in the course of this research. This paper presents the main difficulties encountered by the scholars during the compilation process. The results of this study add to previous research in the field of terrorist propaganda archives. Further research in the area is needed for a more complete understanding of this specific kind of compilation in a context of online global censorship.

Based on second half of the 20th Century television archives and early 21th centuries webarchives from Europe, this historical exploration aims to retrace the ways in which advice was given and recipes suggested for preparing meals. From a broader perspective, the paper lies at the crossroads of a history of food and a history of health and nutrition. The paper suggests a methodology for analyzing the transformations of audiovisual presentations of recipes from the 1960s television to early 21st Century web videos, via a two step web crawl. How have cooking instructions, and with them the act of preparing food, been transformed between the hertzian world of television and the digital recipes of the web? In exemplary manner we study the transformations of a specific time transcending recipe and its representations: the French pancake (crêpe) recipe. What do these transformations tell us about our relationship to food, to preparing meals and about the web as a tool for information and communication in the private sphere represented by the kitchen and the actions that take place there? Beyond the comparative web-video and television approach, our contribution widens its methodological perspective questioning how a prepared cold web crawl relates in terms of archival sources to prior oral history archives that created in a similar fashion archives for historians that were sought and created rather than institutionally gathered thereby transforming the “historian's craft”.

This article aims to uncover some of the challenges researchers face when collecting information from community spaces accessible on the Web. A case study of an investigation conducted in 2020 on a private Facebook group allows us to revisit the process of constructing an archive of such a space. We particularly examine the variations observed between data from Facebook's servers and those collected through an ad hoc system. Three examples of data variations are explored: the intrinsic coherence of the data provided by Facebook, the discrepancies between data directly from the platform and those gathered by an ad hoc collection system, and finally, the gaps observed between the narratives present on the platform and the quantitative data collected. In conclusion, the article advocates for combining information sources when conducting digital fieldwork.

The Harlem Shake phenomenon went viral in February 2013. It was remixed and reenacted by thousands of people, including in France. It combined characteristics of viral phenomena such as the international, cross-platform, cross-cultural, and cross-linguistic dimensions, as well as being both extravagant and ephemeral. As part of the Hivi project (2021-24), dedicated to the history of online virality, and the BUZZ-F project, which focused specifically on viral phenomena in France and was conducted with the support of the DataLab at the National Library of France (BnF) during 2021-2022, it served as a prominent example of virality and a case study for exploring and analyzing its preservation and searchability in web archives. This article reflects on the collaborative work between researchers and web archivists to identify the Harlem Shake within the BnF web archives, create and document useful and representative datasets, contextualize them, and offer visualizations. It highlights the contributions as well as the biases and limits identified, the research strategies that were implemented (through URLs, calculating virality based on the collection history and the interpretive challenges, etc.). Finally, it considers the opportunities this may open up for research and web archiving.

This paper highlights how archiving the literary web requires an epistemological stance that involves a reevaluation of traditional concepts in literary studies: work, author, writing (etc.) The paradigm shift brought about by digitally born and editorialised literature affects the way we build, read and interpret the literary web archive. Using a case study that we will call the “Marilyn Project”, led by Anne Savelli, we will show how the constitution of a “wild” corpus by researchers (turned for the occasion into self-taught archivists) reveals a blind spot in institutional archiving projects. This blind spot can be explained not only by the contemporaneity of the corpus, but above all by a conceptual displacement of literary phenomena in digital space.

This article proposes a methodology for building a corpus and archive of the French-language literary web. It borrows from both corpus linguistics and web archiving, and proposes an original method for acquiring accurate data from the web. Indeed, the fundamental problem in acquiring this corpus lies in the difficulty of identifying web literary production in all its diversity: web literature is not directly identifiable because we have neither keywords nor specific or recurring indices. In this corpus, we are therefore using a variety of complementary methods and tools, which, through the complementarity of viewpoints and always under manual control, enable us to build up a corpus that is, if not exhaustive, at least representative of this literature. To this end, we have borrowed concepts such as the sociability network, and set up linked data sets to describe the structure of these writers' communities, based on the various phenomena of mutual recognition. Finally, beyond this methodology and its validation, we present a few elements relating to corpus structuring, and in particular its indexing.

This article proposes a methodology for building a corpus and archive of the French-language literary web. It borrows from both corpus linguistics and web archiving, and proposes an original method for acquiring accurate data from the web. Indeed, the fundamental problem in acquiring this corpus lies in the difficulty of identifying web literary production in all its diversity: web literature is not directly identifiable because we have neither keywords nor specific or recurring indices. In this corpus, we are therefore using a variety of complementary methods and tools, which, through the complementarity of viewpoints and always under manual control, enable us to build up a corpus that is, if not exhaustive, at least representative of this literature. To this end, we have borrowed concepts such as the sociability network, and set up linked data sets to describe the structure of these writers' communities, based on the various phenomena of mutual recognition. Finally, beyond this methodology and its validation, we present a few elements relating to corpus structuring, and in particular its indexing.

Digital press archives - electronic press or paper press - fully contribute to freedom of expression and information and must, as such, benefit from legal guarantees as to their integrity. However, their switch to the Internet proves problematic with regard to the rights of individuals, due to the new temporality of the provision of information. The permanence of information, and its possible reappearance, thanks to search engines, can be sources of harm for people. It is obviously not possible, due to freedom of the press and the rights of history, to delete the disputed information from the original article because the intangibility of this particular corpus must be preserved. The judges guarantee this intangibility by rejecting any right to erase press information, a principle with which the personal data protection system is imbued by a set of exemptions. However, it is appropriate to prevent, in certain cases, this information from resurfacing again, hence the need for modulated accessibility to these archives. It is therefore possible to carry out certain dereferencing while anonymization must remain an exceptional solution.

This article looks at the uses and expectations that researchers in the social sciences and humanities may have of web archives. Based on research carried out as part of the ResPaDon project, we carried out a survey of players in higher education and research. After putting the survey lead by the national library of France (BnF) in 2010 into perspective, we will analyse the changes in usage and the attitudes of the respondents. While web archives are now increasingly used and accepted as research material, many questions remain unanswered in terms of methodology, technique and researcher training. Between “do-it-yourself” and institutional collaboration, we will present a diversity of configurations and the associated issues.