DHQ: Digital Humanities Quarterly
2017
Volume 11 Number 2
2017 11.2  |  XML |  Discuss ( Comments )

To Visualize Past Communities: A Solution from Contemporary Practices in the Industry for the Digital Humanities

Gérald Péoux <GPEOUX_at_U-PARIS10_dot_FR>, Université Paris Ouest Nanterre La Défense & Institut d'Histoire Moderne et Contemporaine (CNRS)

Abstract

In this article, we discuss a way for visualizing interactions or solidarities in a past community. The method that we expound is derived from process-oriented views used in the industry. Therefore, this attempt to adopt practices from outside of the traditional humanities framework must be clearly justified in terms of epistemological position and in terms of choice among the numerous available tools. First, we try to define what the process term could mean in the humanities; then we explain our choice of a representation; we also carry out tests of our method on selected case studies to visualize and question scholarly works. Last, we offer the reader a prototype that must be improved to produce automatic visualization of particular situations.

Introduction

Designing tools that help visualize scholarly works and offer various perspectives is undoubtedly one of the key features of the Digital Humanities. Indeed, as explained by members of the Republic of Letters program, visualizing helps see things “we did not see before” and “get a better sense of the overall shape and structure”.[1]
In this work, we discuss a “method of representation”[2] that provides both a pedagogical assistance and a synthetic tool for such visualizations. Indeed, we propose to explore a core aspect in human organizations, namely the way they set up, consciously or not, a process to structure their practices. In order to highlight and characterize certain practices among others as a process, we develop a rigorous “representation” that we fully expand with the help of the latest web standards. In addition, we connect this representation to a model of data open to any results of the same type. The conclusions that we present here come from encounters between scholars and experts in the quality of industrial processes. From the discussions between these counterparts, finding a common language ideally based on clear rules, in order to ensure a thorough understanding of each other, has emerged as a necessity. What follows is thus intended for pedagogical purposes, but it could also be regarded as a help to put new questions to the foreground.
At first view there seems to be a tremendous distance between research on history and the search for optimal processes that would minimize the ratios of breakdowns and subsequently provide the high level of quality of an industrial production. Yet, our will to learn from the work of one another has convinced us that the notion of process was already present at various periods of time in history. Since in the industrial context people have a long-term experience on how to represent dynamic process-oriented situations from a start point to an end point, we argue that, by means of a proper translation, it is then possible to offer new perspectives on historical works, as long as they contain one or more process dimensions. Furthermore, we assert that these perspectives pertain to the Digital Humanities domain. Indeed, as Daniel Cohen says in his book Digital History, the digital in history mainly points to a representation of historical materials and investigations with the help of “new” technologies [Cohen and Rosenweig 2006]. To expound our point, we expressed scholarship evidence in a process-oriented way to bring it into Digital Humanities material. We chose the production of knowledge as a first field for our methodological investigation to show its pedagogical virtues. In this contribution, we assumed that the case studies we present may be regarded as a set of more or less ordered, more or less planned, more or less tacit processes.
First, we would like to justify that assumption by reviewing works carried out by historians and by showing that they could be visualized in terms of process. Then we shall present some process-oriented representations and we will explain the choice of one of them in particular. We will try to apply our methodology to two particular works, dealing for the first one with information management and for the second one with the industrial production of fabric. Last, dealing with technicalities, we will show a way to translate our representation in a technical fashion with the help of the last web standards and by presenting a client-side prototype based on a specialized JavaScript library.

Process in the humanities

In this work, we define a process as a set of activities bringing about a result, performed with accuracy and full control by actors who may interact with each other as the links of the same chain. This chain of connected activities forms in a long or short period of time, consciously or not. All actors of such a process pursue a common objective which implies solidarities and sociability. Therefore, we argue that if we manage to disclose such processes from the past, we obtain information about the way communities were organized. Thus, our purpose in this article is to propose means and rigorous methodologies to visualize such processes in history. We chose to rest upon works of historians to illustrate our method, although we could also apply it to primary sources. Indeed, once we have identified relevant processes to highlight interactions between actors, we are ready to visualize them with the help of our tool and subsequently to question our sources on the basis on what was revealed by that new display.
First, let us exemplify our point with cases that were studied by historians and that are related to the production of knowledge.
Many works of historians deal with the production of knowledge at various periods (Antiquity, Middle Ages, Renaissance, Enlightenment…) and in different areas (Europe, the Islamic world, China…). In this section, we suggest that some scholarly works could be represented in terms of process if we can agree to a definition that would be both wide enough to model different situations and sufficiently accurate to take into account the specificity of the humanities. In a book published in 2000, Tiziano Dorandi analyzes various aspects of writing in Greek and Roman Antiquity from the draft to the book edition [Dorandi 2000]. In particular, he examines the conditioning of papyrus, the format of its sale (roll or sheet), and the subsequent impact on the writing of the Ancients. Then, he tries to describe the composition of a treatise: from the gathering of documentation to the note-taking in order to build a collection of notes, the author reviews practices of famous ancient writers, such as Cicero and Pliny the Elder. For instance, he indicates that Pliny was used to first making his assistants take notes on wax tablets before dictating his commentaries on a new physical support from well-ordered notes. He also envisages the passage from the sketch to the note-taking and finally to the edition of a book and its diffusion towards various readers. Dorandi’s well documented study describes the production of knowledge by ancient writers grounded on robust ways of making. This invites regarding that production in terms of process, if we consider such as an accurate, very often repeated and constantly improved manner of achieving a particular task.
In an article entitled “Notes on Medieval Production and Valuation in Late-Medieval Brittany”, Diane Booton investigates the different steps followed by a priest in early fourteenth century Brittany to produce a missal and a psalter [Booton 2006]. The study is carried out on the basis of “tax registers, ecclesiastical payment ledgers and municipal account books.” From these sources, the author identifies several groups of actors who might have participated in the elaboration of the work: parchment makers, papermakers, scribes, illuminators and bookbinders. For each of these groups, the author gives precise information about its internal organization and its interaction with its social and economic environment. For example, we learn that the wages paid to skilled and experienced scribes represented a great amount in the total cost of the production of a manuscript. We understand that the scribes were organized as a team led by a coordinator who might also be in charge of “contracting with a music annotator, illuminator, bookbinder.” If we see this situation in terms of process, we can assert that a process is something which has the property to be connected. Indeed, the different groups involved in the making of the psalter must achieve their own task and forward the result to the next group. This is of course a sequential view of the chain of processes which does not forbid that certain parts of the manufacturing can be led in parallel (for instance, the making of the cover and the copying of the manuscript).
Other works of historians dealing with the production of knowledge in other cultural areas could also be quoted. As an example, let us mention the study of Johannes Kurz who proposed in 2007 an article about “The Compilation and Publication of the Taiping yulan and the Cefu yuangui”, in which he analyzes the constitution of major works (encyclopedias and digests) in the Song period (late tenth century), and presents the personnel who was to take part to it under the command of the emperor [Kurz 2007]. This article also alludes to the collection and selection of the sources that prevailed in their fabrication.
These foregoing examples show that a number of human activities can be modelled by a set of processes and from this point of view, it allows us to emphasize the relationships between individuals in a historical, social and technical context. Hence, our problem is to find a way of representing such processes in a Digital Humanities framework. In this view, let us cite some process representations that have already been tested and experienced in DH.
Many Digital Humanities programs were initiated in recent years, of which some highlighted connections between actors, geographical areas or events and the underlying processes. For instance, the Republic of Letters (already mentioned) consists in drawing up a mapping of the correspondence between Renaissance scholars.[3] This program is strongly related to the social network analysis field which is a relevant perspective for studying a community insofar as it accounts for human relations and consequently the processes that govern their establishment.[4]
Another worthwhile example is the Macroanalysis book by Matthew Jockers in which the author applies digital tools to massive digital libraries data and metadata, such as Internet Archive and HathiTrust, to dig up beyond the layers of literary well-known published works and account for “unjustly neglected” authors. Jockers determines the frequencies of certain words, works on patterns and tabulates his results by genre, date, nationality or origin of the authors. This work is inscribed in the “distant reading” program as defined by Franco Moretti in his Graphs, Maps, Trees [Moretti 2005]. In the chapter dedicated to Trees, Moretti rests on Darwin’s Origin of Species, and specifically on the Divergence of Character section to analogously address the issue of literary survival. He chooses British detective fiction as a field of investigation and builds a tree to distinguish the novels that use clues from others, which enables him to highlight that the latter ones were all forgotten. In the second level of branching, he refines the character of divergence and classifies novels with necessary clues and not necessary etc… This tree accounts for literary transformation under the “pressure of social selection”. As in Darwinian evolution, that transformation seems to be the result of a long-term process of which the causal factors must be identified and analyzed. In that case, the “process term” refers to a “natural” evolution not deliberately ordered nor commanded by a community of human beings. For that reason, the tree diagram acts as a classification scheme according to a certain criterion, that is, the divergence of a character. It represents the outcome of one or more underlying processes the nature of which we do not know. But, this tree diagram describes in no way a process in action. Indeed, mapping relations between Renaissance scholars or building tree diagrams such as above reports the macroscopic dimensions of human and social organizations.
In contrast, in this essay, we intend to concentrate on the processes that human communities set up for a precise objective and that dynamically account for an action being made. In this way, what we propose can complement the tools and methods that we have just mentioned above (such as tree diagrams or network mapping) insofar as we focus on the process itself in such a way that we can disclose the atomic dimensions of the situations we investigate. Thus, we consider interactions between elements that may be either individuals or material resources or even more abstract entities such as activities. The choice of such interactions enables us to account for complex situations (behind which we find individuals, organizations or external constraints) that we describe with an identical framework whatever the details that we obtain from historical works.
Therefore, we think that modern methods of process representation could contribute to highlight the interactions between these various elements and, as such, that they would give a relevant description of past collective enterprises like, for example, the production of knowledge.

The choice of a process representation

In order to make a choice of a process representation for visualizing scholarly works, we should first consider what a process means in our contemporary world and especially in the industry where the need for formal and rigorous methods is crucial.
In a modern wording, a process may be defined as a series of linked activities and resources (either material or human) which transform input elements into output ones [Oakes 2013]. Thus, an organization or a collective activity may be modeled as a set of processes evolving in an external environment (represented by constraints that must be respected) focused on a specific objective. The interaction between processes may be regarded as a communication channel in which a flow circulates. It raises the problem of the relevancy of these modern concepts for representing works in history such as those briefly described above. In fact, we do not mean to represent the past world but only “to construct an interpretative framework” for accounting situations described in scholarly works. This “epistemological” position is clearly related to the status of modeling in the humanities. Following Michael Gavin, we argue that our process-based model “do[es] not represent the past, but our ideas about the past” and, in our case, ideas derived from scholarly works [Gavin 2014]. Therefore, when we talk about “representation”, we mean a way of representing the processes that constitute our “interpretative framework”. As a matter of fact, modeling in the humanities does not appear as a novel activity: for example, using “categories to delimit a conception of the world” is already an activity of modeling [Gavin 2014] [McCarty 2005].
There are several ways of representing processes. All of them correspond to different points of view. We can quote, for instance, those focused on heuristic conceptions, usually called mind maps. They are very relevant to account for decision processes but do not seem to correctly represent situations and activities already well experienced as those we would like to illustrate here.[5]
The Unified Modelling Language (UML),[6] mainly related to software development, specifies several diagrams dedicated to precise situations (static, dynamic, sequential, interactive...). In particular, the UML workgroup recommends to the user the Activity Diagram in order to highlight the chaining of activities of a system. It is useful to model an interactive process. More precisely its features are close to those of a flow chart or of an algorithmic tree. Figure 1 is an example of an activity diagram which accounts for a course registration process in an academic institution. Each rounded rectangle represents an activity, that is to say, a step of the process, whereas the rhombus-shaped figure stands for a decision split on the basis of an exclusive choice (an XOR operator in Boolean terms). The grey bars symbolize synchronization steps while the black thin arrows play the role of input or output elements.
Figure 1. 
Example of a UML Activity Diagram
Although UML focuses on software and system designs, it can be extended to other purposes.[7] For example, a more business oriented approach has been standardized by the Object Management Group as the Business Process Model and Notation (BPMN) to provide “the capability of understanding the internal business procedures”.[8] As the UML Activity Diagram notation, BPMN offers several components to model a process oriented view in a business approach. At least, we can cite the event, activity, gateway and connection elements. The event elements (represented by a circle) denote something that happens: the Start event is the process trigger whereas the End event is the result of the process. The activity elements (rounded corner rectangles) describe the work that must be done. The gateway elements (diamond shapes) determine forking and merging of paths depending on conditions (exclusive, event evaluation, inclusive…). All these elements are linked by connection elements (sequence, message passing, and association) to describe the overall process.[9] The UML and BPMN methodologies, which share common roots (as they are standardized by the same consortium), appear as good candidates for our subject.
As a matter of fact, we preferred a methodology of representation derived from the quality of process and we will try to justify our choice further. The quality of process ensures that, for a certain probability or more precisely at a certain error rate, which must be minimized, a process will produce the expected output element. The philosophy of such a representation consists first in observing what already exists and then, upon these observations, in establishing a diagnosis of what must be improved. For this purpose, such methods provide robust ways of building a framework or a model for apprehending reality. With this process representation, we can position ourselves as observers of experienced situations and not as designers of a system, which we cannot pretend to be when scrutinizing the past. In the industry, minimizing an error rate may be characterized as a “progress”. It is achieved by adopting a “broad vision of the company” that enables obtaining more control of its activities by a fine analysis. This “business process approach” is of course rooted in our contemporary world and we must maintain a necessary critical distance if we wish to adapt it to our subject. And to do so, we must first state precisely that, in our case, “progress” cannot be related to any “improvement” in terms of quality, for we do not intend to make an audit of the past scholars. We only wish to set in motion a current methodology of quality improvement to visualize past activities. In that perspective, the “broad vision” offered by that methodology meets the expectations of historians using Digital Humanities techniques for it implies a fine and rigorous analysis of all related activities.
Let us now briefly describe the methodology. Companies implement process referential and maps, which define the interactions between processes, show and reveal the overall coherence. It is a tool of communication and therefore it represents the synoptic view of all business processes (sequence and interactions between processes).
Processes and roles relating to these are generally independent of the organization and functions. The method of process representation that we consider here is inscribed in an overall toolbox usually declined in five phases of which the main and key objective is to converge to a “zero-defect” approach in a chain of production.[10] To that purpose, the expert must build in the first phase a process map of the production, and then analyze it in order to improve and optimize the quality of making. We would like to focus on the first phase of the toolbox that proposes a robust set of tools to establish such a map.
In the reality of industry, the project manager must ensure that basic questions such as “the project relationship with the overall objectives”, “the nature of the problem” and the “goal” have been understood by all stakeholders. Hence, it follows a progressive approach, including multiple steps, each of them being focused on a specific problematic.[11]
During the so-called first phase, the project manager starts by understanding the current process and its related characteristics (process boundaries, the way products and services are actually delivered, etc.). For that, several tools are useful to visualize and point to the process that needs improvement, to identify start and stop points of the process under consideration, to identify the customers of the process and to ensure a shared understanding of key elements of the process, including what it looks like and what it should deliver. These tools are the “SIPOC”, the “high-level process maps” and the “detailed process maps”.[12]
“SIPOC” stands for Supplier, Input, Process, Output, and Customer. It deals with business process representation as the logical organization of suppliers, inputs, and work activities designed to produce a required end result (product or service output) for internal or external customers.
Then, the “high-level process maps” and “detailed process maps” representations propose a focus on the process itself detailing its activities, main flows, and other elements.
The SIPOC, as a first representation, greatly helps to understand the process context. This is very important, as chains of process in companies are quite usual, meaning that the outputs of the first become the main inputs of the next one, etc. To be brief, we can state that, in the SIPOC wording, Supplier provides inputs to the process, Input spans materials, resources, and data necessary to execute the process; Process is a collection of activities that accepts one or more inputs and creates output that is of value to customers; Output is related to products, services or data resulting from a process whereas Customer points to internal or external entity that receives outputs.
When focusing on the process itself, the “high-level process maps” and the “detailed process maps” representations are useful. An example of a process map that we will use and detail further in the present article, is the flowchart. A process flowchart includes several key elements which are the main activities and milestones of the process; the flows between activities, e.g. main inputs and outputs; the roles associated with the various activities; the different elements supporting the run of a given activity.
Figure 2 is an example of the flowchart template used to represent the various processes in the Thales Group Chorus 2.0 referential. Table 1 is the associated caption.
Figure 2. 
“High-level process maps, flowchart principle” 1/2
The role(s) associated with an activity. The person responsible has to be highlighted (bold, italic, underline…). Each activity, including the decisional milestone, has to be associated with one or several roles.
Activity title: Short (50 characters max), typically using an “action verb” and a complement, “A few keywords” as typically a set of substantives allowing to understand the activity main content.
Case where, for this activity, a lower level description is available (flowchart and procedure).
Decisional milestone title: Short (50 characters max), typically using an “action verb” and a complement.
Document (instruction, guide, template, form, manual, example) or folder supporting the activity execution.
Indicates for each activity, decisional milestone, the main associated input(s).
Indicates for each activity, decisional milestone, the main associated output(s).
Table 1. 
“High-level process maps, flowchart principle” 2/2
To summarize our choice, we state that we intend to apply a method used in the industry to describe and to visualize particular situations reported in scholarly works. This method consists in coupling a generic and macroscopic representation (SIPOC) with a more detailed process map (namely a flowchart) specifically developed and thought of by the Thales Company.[13] The main benefit of this flowchart on BPMN is its ability to make visible, in a simple way and on the same diagram, the roles (namely the main actors and the stakeholders), the activities with their inputs and outputs and the implemented means. In addition, the flowchart is situated at a rather satisfactory level of accurateness but it is also sufficiently macroscopic to provide an overall view that corresponds to the expectations of any Digital Humanities practitioner whose objective is to comprehensively render precise results. In addition to its intrinsic simplicity due to the vertical layout, all these reasons inclined us to implement this representation in the framework of Digital Humanities.

Case studies: application of the SIPOC-Flowchart representation

Indeed, in their work on Slavery in American Society before the Civil War conducted at the University of Virginia, W. G. Thomas and E. L. Ayers argue that one of the activities of a Digital Humanities project is the translation of “professional scholarship-evidence into forms that take advantage of the possibilities of electronic media”  [Thomas and Ayers 2014].[14] This translation may be achieved by a relevant arrangement of the information system. In a basic information system architecture, we usually identify several components. Among them, we find the data model which organizes the interrelated pieces of information (usually in a database) and the “view” which displays results extracted from that database according to a query. The view component has gained huge importance so that it usually influences the design of the data model.[15] Figure 4 represents a schematic articulation of the main components for our Digital Humanities pattern.
Figure 3. 
Three main components based on the Model-View-Controller pattern. In our solution the View component is a SIPOC/flowchart. The Model component is discussed below whereas the Controller is a piece of software which extracts data and prepares them for the View component.
In this section, we shall see how our process-oriented methodology can offer the opportunity to build a view component with the intention to visualize historical data.
We shall exemplify our methodology on case studies drawn from Ann Blair’s Too much, to know [Blair 2010] and we will try to extend it to another scholarly work.[16]
We first chose Too much to know because it embraces most of the aspects related to the production of knowledge in a particular period, that is, the Renaissance. Ann Blair’s book is indeed about scholarly information management before the modern age. The author devotes five well-documented chapters on that question. In particular, in chapters 2 and 4, she gives a lot of insights on the note-taking activity regarded as a central step towards information management (chapter 2) and on compiling, which can be viewed as a following step resting on the output collected from the note-takers (chapter 4). These chapters are documented by numerous and detailed examples which give hints on how the different actors could manage the abundance of information. We would like to reread these chapters from the perspective of our methodology, first by drawing a generic SIPOC scheme. The main objective of this generic scheme will be to offer original views of historical data and, probably in a further work, to build an atlas of processes set up by early modern scholars. Its only virtue will be that of positioning the different activities, as parts of the process, in the global process that leads to a final tool of information management. On this basis, we will examine a case study for which some of the previously detailed phases are consistent with historical materials.

Generic SIPOC

In chapter 2, the author sets the bases of a history of note-taking, as “various kinds of writing in response to listening, reading, or thinking…for the production of a composition or report”  [Blair 2010, 64] pointing to the difficulty of such a history due to the loss of material or its immediate destruction inherent in its use. Indeed, as mentioned, “erasable writing surfaces were commonly used for temporary notes”  [Blair 2010, 65].These surfaces were wax tablets bound in a codex form, pocket-sized writing tablets, slate blackboard, sand tray…In addition to the fragility of those writing surfaces, the discarding of notes, which was common among scholars and especially their heirs, is another feature that makes the taking-note process difficult to analyze [Blair 2010, 66]. Another reason which accounts for the loss of notes is their “physical integration into a manuscript used for printing” as illustrated through Pierre Bayle’s Dictionnaire historique et critique [Blair 2010, chap.4]. We can already anticipate that “taking notes”, “discarding notes” and “integrating notes” will be categorized as activities in the SIPOC representation whereas the so-called associated means will be represented by all kinds of “erasable surfaces”: on one hand wax tables, sand tray, slate blackboard as mentioned above, and on the other hand “loose sheets stored in bundles” and notebooks as stipulated further. What could be the roles corresponding to those activities in the SIPOC flowchart? If we read Ann Blair more thoroughly, we learn that the notes taken by readers on erasable surfaces were often destroyed. Those “first-order” notes, as characterized by the author, served sometimes as a basis to “second-order” notes put on a more durable surface. These could also be put into circulation after revision and authorization by the speaker [Blair 2010, 64–65]. So, the listeners (students, secretaries…) in conjunction with the speakers may be considered “roles” in the “first-order” and/or “second-order” notes production activities. In some cases, the author gives details on how scholars took notes or made annotations when reading a book: sometimes they highlighted passages with keywords in the margins to produce a running index [Blair 2010, 71]. One feature relevant to the process-oriented note-taking activity is the debate between scholars who advocated reading without interruption and those who, in contrast, recommended stopping reading in order to excerpt and to copy passages in a notebook [Blair 2010, 72]. Note-taking most often took one of two forms: either “by epitome or abridgment,” that is, by paraphrasing or summarizing the sources, or “by head or commonplaces,” that is, by copying passages from the source into a notebook under a commonplace heading. These ways of working shed light on the “note-taking process” that we will link to the “first-order” notes activity.
As the quantity of notes increased, it became of central importance to set up tools for their management. One crucial tool was the assignment of headings to a set of notes. As Blair mentions, the choice of such headings was rarely discussed by scholars, although it was “crucial to effective note-taking” and required judgment, as part of a decision process [Blair 2010, 88]. Another important decision involved the number of headings to be used: to be useful, headings had to be chosen accurately and their number had to be great enough to guarantee sufficient subdivisions, but small enough to prevent the storing of too many notes under the same topic. Another tool for managing notes was their indexation which supposed that each passage of a notebook “would be entered into the index under a keyword”  [Blair 2010, 91]. One way to do so was to alphabetize the passage “under its principal thing” and to draw up an index on “sheets corresponding to each letter of the alphabet” with references to the notebooks. That way, the index could “receive new headings over time”. This indexation of a notebook is therefore described as a living process which, together with the constitution of headings, corresponds to an activity in a SIPOC that could be entitled “notebook management tool elaboration”.
What was the main purpose of excerpting and accumulating notes in notebooks? Blair tackles that question further in chapter 2. The historical materials she presents in her development depict the notes taken by scholars as “an aid to writing”, as worded by a subtitle of the chapter. In our view, that “aid to writing” is one branch of the process of writing, and may be considered one of the steps that describe, in the SIPOC, the achievement of a reference book in early modern times.
In chapter 4, more focused on the compiling activity, the author gives many hints on the methods used by compilers and on the social and technical context they were acting in. In particular, she discusses some social and legal elements, which, to our view, should be included in the process description by a SIPOC. For instance, the questions of obtaining privileges, which were only valid within a certain jurisdiction [Blair 2010, 179], or of avoiding censorship [Blair 2010, 204] were determinant in the making of a reference book, although not purely technical. In figure 4, we propose to represent the main steps of the reference book creation process, as an application of the SIPOC diagrams presented in the previous section. In the SIPOC symbolism, the blue frames represent activities that may be described by another flow chart, that is, that have a degree of complexity which justifies a detailed process representation.
Figure 4. 
Global flow chart delineating the compilation of a reference book creation process. At this level: each activity may be described by another flow chart which plays the role of a focus on it.
Regarding the compiling activity itself, it may be considered the result of collaborations either diachronic, when “compilers relied on earlier work”  [Blair 2010, 208], or synchronic, when compilers relied on amanuenses [Blair 2010, 209]. So, earlier works such as notebooks, or aids from helpers, may be categorized as inputs in our SIPOC-based reading.
If we look more precisely at the technical activities as practiced by the actors, we find some rather basic ones, such as “selecting a passage from a source and assigning the passage to a topical heading”  [Blair 2010, 210], but also some new techniques based on “the use of slips and of cutting and pasting from printed books.”
Thus, to describe an overall process, we can consider flow charts at several levels: at the top level, we represent global processes whereas more detailed processes may be specified at a second level by another flow chart.
This is described in figure 5 where we translated the note-taking activity according to the elements of Blair’s work previously reported.
Figure 5. 
A focus on the note-taking activity.
As we can see in figure 5, each activity is still framed by a blue rectangle which means that we consider possible another level of accurateness. Below, we exemplify that third level flow chart with Sacchini’s “first-order-notes taking” activity.

A case study

This foregoing review of Ann Blair’s chapters 2 and 4, translated into a process oriented representation, gave clarifications on the sequences of the activities (at different levels) that yielded a final intellectual product corresponding to the expectations of scholars in early modern Europe. But this generic representation based on SIPOC and flow charts only assigns a correct position to each component in the global process (roles, activities and associated means) provided that the component is effective in the process. If we focus now on a particular case with the intent of building its relevant SIPOC, we should be able to visualize its shift from the generic representation. In particular, we will inevitably have to contend with gaps in historical information, which amounts to the absence of certain components in the SIPOC flowchart. The choice of a particular case could be made in function of different criteria. For instance, we could follow the process of one scholar or a group of scholars who committed themselves to a particular production. That kind of process would spread over a certain geographic area, over a short period of time (that of the scholar’s career). In contrast, we could also take into account the achievement of a book over a long period of time, such as Theodor Zwinger’s Theatrum Humanae Vitae, which, as related in chapter 2, spanned over almost one century and a half, or the Polyanthea produced during 178 years.
For the sake of “simplicity” (if the term is relevant for such a study), we propose to focus on the Jesuit Francesco Sacchini as a note-taker. Sacchini’s way of taking notes contrasted with that of others as he recommended “the interruption in reading that resulted from stopping to copy a passage into one’s notebook: it slowed down reading and aided retention”  [Blair 2010, 72]. In more detail, we learn that Sacchini recommended taking notes in two steps: first he selected passages and wrote them out on a notebook “in the order in which they were encountered”, and second he copied the passages out in another notebook under commonplace headings [Blair 2010, 77]. If we read Blair’s account correctly, we assume that Sacchini read a passage once and wrote it twice as just mentioned.[17] The selection process was materially performed by underlining the passage of interest in a book [Blair 2010, 86]. Then, as we pointed out in the previous section, the choice of a heading was also an important step in the process. Concerning Sacchini’s method of heading choice, Blair alludes to it [Blair 2010, 89], but it is likely that we do not have sufficient materials to give a thorough development. It is anyway worth mentioning and positioning it in the flow chart, even if we are not able to provide a focus on it. It is likely that the choice of a heading was made when copying the passage out for the second time, enhanced by familiarity with the text acquired during the first round of note-taking. That iterative process is described by figure 6.
Figure 6. 
First-order-notes taking exemplified by Sacchini’s method. In this third level flow chart, each activity is framed by a black rectangle which means that we do not intend or that we are not able to give more details of the process.
In this section, we showed that the process oriented view, which is the SIPOC diagram coupled with flow charts at different levels, proved to be a powerful candidate to visualize and represent data at the front-end of a digital system where historical work must be stored. Beyond its advantages in terms of visualization, it compels an accurate and rigorous analysis of activities, and in so doing, it enables setting up a powerful focus on specific cases. Therefore, it provides the scholars with a tool reference for comparing different cases and for raising subsequent questions. In addition, that tool does not concentrate only on individuals, but it defines, in a broad meaning, activities and roles related to each other, which gives an account of the interactions between the actors and their environment. Last, one more advantage of such a methodology lies in its ability to provide a global and overall representation of a past enterprise.
However, we must be careful when using this methodology. Indeed, the first problem we encounter when we try to apply the SIPOC and flow chart representation is related to the effort we must make to deeply understand all its subtleties. Indeed, this methodology was not first designed for historical works: it is then necessary to adjust it to the needs of historians (for instance, by redefining some of the vocabulary). Furthermore, it rapidly proves to be inoperative if the amount of scholarship evidence is not sufficient, as it is demanding many details to be enlightening.

Another case study: the industrial production of fabric

This foregoing and detailed part has shown that the SIPOC-flowchart representation is well-adapted to a certain type of scholarly work, namely any work related to the production of knowledge. But, we can now legitimately examine whether that method of visualization could be applied to other works. To answer this problem, we favored a pragmatic approach: as our point is not to build a theoretical framework within which we would demonstrate logically the validity and the relevancy of our method for a general case, but rather to show that it can help visualize results of selected scholarly works in a thorough way. To that end, we put our attention on a recent publication, “Copy and Counterfeit in the Renaissance” [Duclos-Mounier and Nativel 2014], which collects the proceedings of an international conference dedicated to this topic. It is a 500-page volume divided into two main parts: the first one is theoretical and deals with art, book, or science and religion history; the second one, more practical, investigates different cases of literature, book, art, science, religion or economy counterfeit. One of these case studies describes and analyses “textile copy and counterfeit in the Renaissance”  [Jeggle 2014]. In this chapter, the author makes the distinction between imitation and copy or counterfeit. In certain cases, imitation consists of the substitution of raw materials used to make the textile. For example, the origin and the quality of the fiber used to make the textile were of first importance: making products with wool from another origin than expected (maybe cheaper) was a constant temptation, but these products were usually considered falsified. Another example: were the silk threads made with first choice silk or with scrap material? In Italy, the regulations demanded that thread be spun with pure materials, which implied thorough controls. The next step in textile fabrication was the weaving phase. It was then possible to circumvent the quality standards by introducing threads of different quality levels or by using less thread than recommended. We learn from the chapter that it was a source of numerous conflicts within the production centers. After the weaving, many textiles were bleached or dyed. The quality of the dying was also a matter of discussion: some colors could be made with different materials. Their durability and the damages caused by the subsequent chemical reactions were variable and influenced the quality of the final result. Last, high quality textiles needed fine finishes of which the quality determined also that of the product. During these last phases, there were numerous opportunities for counterfeiting: quality of the textile and quality of the fabrication procedures were the main parameters to play with for copying original products. To prevent these practices, the authorities had organized committees that examined the fabric quality. They had set up production standards and inspection organizations that certified the products and, when necessary, they declared fraud perpetrations. For instance, in Augsburg, woolen sheets were inspected at several times during their production: these inspections mainly involved the material quality of the fabric and also the quality of its treatment. Surely, the practices of inspection and certification were different according to the areas and the production places. The inspection and its subsequent certification were materialized by seals printed on the fabrics with ink, wax or lead. The wax color indicated the quality level.
As the reader can anticipate, this chapter of the book is rich and full of many details that can lead astray someone not familiar with that scholarly topic. After that first general presentation, the author expounds a detailed case study focused on linen production in Westphalia. As our purpose is not to summarize that fascinating study, we only would like to benchmark our method of representation to the previous general considerations.
In figure 7, we tried to render these considerations in a SIPOC-flowchart representation.
Figure 7. 
Flowchart representation of the textile fabric manufacturing.
In this diagram, we chose to define six main activities controlled by four milestones corresponding to the different inspections and certifications that we briefly presented above. This flowchart rests on the principle that a process which would successfully pass all these phases would produce certified and not counterfeited fabric. All these activities are represented as general ones, which means that they may be unfolded into subactivities to provide more details. This could be achieved, for example, by extracting the information contained in the footnotes of the chapter. Note that the bypass arrow between the “Dying quality control” and the “Quality control” milestones models a “normal” fabric, namely a fabric that has not been finely finished. On the right side, we attached to each activity its inputs, outputs and its associated means (in a parchment symbol). Note that the question mark we appended to the associated means state that the author of the chapter did not clearly mention them. However, when we figure out in terms of process the situation that he describes, the question of the means implemented by the textile workers must be raised at the precise place pointed by the flowchart. This is another virtue of such a representation: the gain of an overall view inclines the reader to raise questions that pertain to a specific point and that did not arise the first time.

Technicalities: data model and the SIPOC-flowchart representation as a frontend-view

For more than forty years, digital system architects or designers have adopted common methods to build databases according to standard models. As previously mentioned, an information system, that is to say the core of any Digital Humanities project, usually rests on three main components (see figure 3). First, the data model, mainly a database, contains the scholarship work organized according to a rationale. Then, the view acts as an interactive component between the data model and the end-user. It is a simple web page, a rich interface application (or RIA based on JavaScript framework) or a traditional graphical user interface (java based for example), depending on the level of sophistication. Last, the controller is the component that connects to the database, retrieves the data relevant to the user query and prepares them for a display on the view. The main advantage of such architecture lies in its ability to distribute the overall development among different specialized skills (database design, programming and view web-based design). We have already presented the process map methodology as the view component relevant to certain historical works. Let us now briefly describe what could be another central brick in Digital Humanities: the organization of historical evidence or the design of specific databases. In human sciences, the databases must essentially follow the relational model based on the second and/or third normal forms [Codd 1972]. From this architecture, it is then easy to produce XML or JSON files,[18] provide them with vocabulary or ontologies, implement an efficient search engine to retrieve data and make them capable, through relevant protocols, to open a dialog with other web-based applications and exchange data.[19]
Those technological “novelties” may not be regarded as simple tools by the historians any longer. At least, they offer new visions on the historian works, new ways of analyzing and new ways of researching [Cohen and Rosenweig 2006]. If we focus on our SIPOC approach, we can notice that the foregoing flowcharts are good candidates for segmentation in atomic entities as recommended by the relational second and third normal forms. Indeed, we can anticipate, in the case of a database design, that activities would be modeled by an entity which would cover several attributes such as its title, a description and probably other useful fields. That entity would have a relationship with others: a “role” entity, an “input” entity, an output entity and a document entity. These relationships would be attributed cardinalities whose values would depend on their nature. As it is obvious that a role may be linked to several activities and, reciprocally, an activity may be performed by several roles, the relationship activity-role will be of type “n-m” with a “0-N” cardinality on each side.[20] Concerning the input, output and document entities, we can consider, for the sake of simplicity, 1-n relationships with the activity entity, which means that an activity is related to, at most, one input (respectively output or document) and that the input activity may be linked to another activity. As a consequence of such a choice, the field that describes the input (respectively output or document) must be multivalued in order to account for the fact that, in reality, several inputs may correspond to one activity.[21] The entity-relationship scheme that represents that model is as below:
Figure 8. 
Simple entity-relationship model for the activity of a flow chart. The “id_XXX” stands for the identifier of the entity.
The diagram above must be made more complicated if we want to account for the sequence of the activities, as it appears in the flowchart process. One way to model that sequence consists in assigning a “sequence” entity to the “Role-Activity” relationship. In addition, as seen previously, an activity may be expanded in “subactivities”, which amounts to create a “tree” of activities of which we limit, for clarity reasons, the number of branches to 3. The modeling of such a feature may be a reflexive relationship on the activity entity itself. Therefore, a more definite entity-relationship model might be represented by figure 9:
Figure 9. 
Entity-relationship model for the activity of a flow chart with sequence and expansion in subactivities.
To practically exploit such a model it is necessary to convert it in a relational database (using the well-established rules), then to fill it with data. In order to represent a particular SIPOC, one could develop an engine which would retrieve the relevant data in function of a user request and which would generate an HTML page to display a flowchart. Indeed, the recent HTML5 standard includes a “Scalable Vector Graphics” (SVG) [22] engine that is able to produce image on the fly. Thus, if the engine gets good geometric data calculated from the database, it is then possible to generate flowcharts upon requests through a well-designed interface.
One way to draw automatically a flowchart in SVG format would be to provide the application controller with a converter from relational to JSON format. Having JSON data retrieved on the client side (namely in a browser) enables drawing a flowchart in SVG/HTML format, provided we wrote JavaScript code based on libraries specialized in data visualization such as the Data Driven Documents JavaScript Library (d3js).[23] In order to show the feasibility of our assertion, we remained pragmatic and we developed a quick prototype based on d3js. We assumed that a flowchart could be stored in a JSON array where each element (namely each JavaScript object) stands for an activity and its associated means and roles. Therefore, in a JavaScript section of an HTML page, a flowchart could be represented as below:
var sipoc = [ {"activity":"Notes Taking", "MeanIn":"Books Treatises, erasable
                    surfaces...", "MeanOut":"Various written Surfaces" "Role":"Listeners, Readers"},
                    {"activity":"Notes discarding", "MeanIn":"Notes", "MeanOut":"Nothing or
                    Integrated notes","Role":"Compilers, heirs"}, {"activity":"Printing",
                    "MeanIn":"Notes", "MeanOut":"Reference books" , "Role":"Printers"}, ]; 
This data structure fits to the relational database model seen above and can thus be easily built in a client (browser). Then, in the client side, we only have to perform some appropriate d3js function calls (using the functional property of the JavaScript language) such as:
var svg = d3.select('#graph') .append('svg') .style('width', 2000)
                    .style('height',1000);
to create a SVG plane, and:
svg.selectAll('rect.activity') .data(sipoc) .enter() .append('rect')
                    .attr('class','activity') .attr('x', xBarPosition) .attr('y', function (d,i)
                    {return ((i+1)*yBarPosition);}) .attr({width: widthBar, height: heigthBar,
                    style:"fill:#cb842e;stroke-
                    width:"+strokeArrowActivity+";stroke:#a00c0c"});
to create an activity block.
In figure 10, we reproduced the result obtained automatically by the code given in appendix. For the sake of clarity, we have represented only a few dimensions of the flowchart (in black arrows) relative to the web page edges.
Figure 10. 
Output of the JavaScript/d3js prototype derived from the SIPOC-flowchart process representation.

Conclusion

In this article, we proposed a solution derived from a combination of several best practices in order to account for scholarship evidence of professional historical works. The best practices we are pointing at come either from the industrial management process, from the software development and from the critical distance that we must take to produce a relevant visualization of a scholarly work. Concerning this latter point, we carried out tests of our solution on two different topics that were objects of investigations by historians, namely the production of knowledge and the question of counterfeiting in the Renaissance. We saw that the SIPOC-flowchart representation gives an overall view of a case study that resulted from a scholarly work. It gives an insight to the shape and structure that constrain the practices of a community. But we also argued that building a SIPOC-flowchart helped us raise problems and questions that we did not anticipate before, in the sense that our representation suggests (through the blanks that it left) solving and answering particular problems.
Last, we think that it is very likely that any other work involving process or, more generally speaking, solidarity between different actors in a specific social context would be easily translated into the Digital Humanities field by our solution. Indeed and in a practical way, once we have properly designed a database to represent data coming from historical works, we open the door to the web ecosystem and to all the features it offers. In particular, we can imagine, following Moretti’s call for a worldwide collaboration [Moretti 2005], research programs that would involve scholars in collaborative relationships for modeling and providing data in a process-oriented view as we discussed above.

APPENDIX: PROTOTYPE FOR AN AUTOMATIC FLOWCHART GENERATION

Notes

[2]  In fact, we will qualify this term further on.
[4]  This work is not the place to review all the literature on social network analysis inasmuch as it is extensive. Let us only mention [Scott 2012].
[6]  See http://www.uml.org.
[7]  For example, the SysML modeling language refers to engineering systems.
[8]  There are several other notations. Let us mention, for example, the workflow language Yet Another Workflow Language (YAWL) (see http://yawlfoundation.org).
[10]  This methodology is called “Define Measure Analysis Improve Control” (DMAIC); see [Shankar 2009], [Rever 2013] and [Oakes 2013].
[11]  We will also see below in this article that this approach can be a source of rich perspectives, offering the possibility to emphasize and compare various process evolutions.
[12]  For a synthetic presentation of the SIPOC process representation, see [Oakes 2013], see also [Shankar 2009].
[13]  In that sense, we follow the recommendation of Harry Rever, the Director of Lean Six Sigma International Institute for Learning: SIPOC makes sense at a high level exercise whereas a more detailed process map is the next logical step. See [Rever 2008].
[14]  In this work, the scholars applied Digital Humanities methods to explore the way slavery divided American society before the Civil War. They used the capabilities of new technologies “both for analysis and for presentation of the argument”, relying on Geographic Information System (GIS) and on Extensible Markup Language (XML) to “connect large amounts” of information.
[15]  See for example the MVC pattern as discussed in [Reenskaug and Coplien 2009].
[16]  Indeed, we will try to examine the conference proceedings of which the topic is “Copy and counterfeit in the Renaissance”, see [Duclos-Mounier and Nativel 2014].
[17]  This is suggested in [Blair 2010, 77] and confirmed in [Blair 2010, 79], and further in [Blair 2010, 86] by an example of a student who followed Sacchini’s method.
[18]  XML stands for eXtended Markup Language (see http://www.w3.org/standards/xml/and http://www.w3schools.com/xml/default.asp), whereas JSON (JavaScript Object Notation) describes a JavaScript Object (see http://www.json.org). Both XML and JSON are intended to data interchange.
[19]  In the field of bibliographic databases, we can cite protocols like OAI-PmH. See http://www.openarchives.org/OAI/openarchivesprotocol.html
[20]  This means that the entity is involved in the relationship at least zero times and at most several times.
[21]  So, the description field of an input element would correspond to several inputs. For instance, “input1” will be described by “first order notes, books, letters”. This is technically simpler than considering three elements “input1” (first order notes), “input2” (books), “input3” (letters) that we should link to the same activity through an “n-m” relationship. It induces more redundancy but offers more simplicity.
[23]  On d3js, see http://d3js.org.

Works Cited

Blair 2010 Blair, A., 2010. Too much to know: managing scholarly information before the modern age , New Haven: Yale University Press.
Booton 2006 Booton, D.E., 2006. Notes on Manuscript Production and Valuation in Late-Medieval Brittany. The Library, 7(2), pp.127–153.
Codd 1972 Codd, E.F., 1972. Further normalization of the data base relational model. Data base systems, pp.33–64.
Cohen and Rosenweig 2006 Cohen, D.J. & Rosenzweig, R., 2006. Digital history: a guide to gathering, preserving, and presenting the past on the Web, University of Pennsylvania Press. Available at: http://chnm.gmu.edu/digitalhistory/.
Dorandi 2000 Dorandi, T., 2000. Le stylet et la tablette: dans le secret des auteurs antiques, Paris: Les Belles Lettres.
Duclos-Mounier and Nativel 2014 Duclos-Mounier, P. & Nativel, C. eds., 2014. Copier et contrefaire à la Renaissance: faux et usage de faux, Paris: H. Champion.
Gavin 2014 Gavin, M., 2014. Agent-Based Modeling and Historical Simulation. Digital Humanities Quarterly, 8(4). Available at: http://www.digitalhumanities.org/dhq/vol/8/4/000195/000195.html [Accessed January 4, 2016].
Jeggle 2014 Jeggle, C., 2014. Copier et contrefaire des textiles au temps de la Renaissance. In P. Duclos-Mounier & C. Nativel, eds. Copier et contrefaire à la Renaissance: faux et usage de faux. Paris: H. Champion, pp. 293–306.
Kurz 2007 Kurz, J.L., 2007. The Compilation and Publication of the Taiping yulan and the Cefu yuangui. Extrême-Orient Extrême-Occident, pp.39–76.
McCarty 2005 McCarty, W., 2005. Humanities Computing, Basingstoke: Palgrave Macmillan.
Moretti 2005 Moretti, F., 2005. Graphs, maps, trees: abstract models for a literary history, London.
Oakes 2013 Oakes, D., 2013. Performance Metrics: The Levers for Process Management, Amer Society for Quality.
Reenskaug and Coplien 2009 Reenskaug, T. & Coplien, J.O., 2009. The DCI Architecture: A New Vision of Object-Oriented Programming. Available at: http://www.artima.com/articles/dci_vision.html [Accessed December 12, 2014].
Rever 2008 Rever, H., 2008. Ask Harry! The Fundamentals of Metrics Part 1: Considerations for “What” to Measure. allPM. Available at: http://www.allpm.com/index.php/our-community/94-article/newsletter-article/411-askharrythefundamentalsofmetricspart1considerationsforwhattomeasurebyharryrevermbacssbbcqmcqcpmp-directorofsixsigmaiil [Accessed January 16, 2016].
Rever 2013 Rever, H., 2013. Applying the DMAIC Steps to Process Improvement Projects: Define-Measure-Analyze-Improve-Control is the “roadmap” to improving processes International Institute for Learning, Available at: http://www.allPM.com.
Scott 2012 Scott, J., 2012. What is Social network analysis?, London: Bloomsbury Academic.
Shankar 2009 Shankar, R., 2009. Process Improvement Using Six Sigma : A DMAIC Guide, Milwaukee: ASQ Quality Press. Available at: http://site.ebrary.com/lib/alltitles/docDetail.action?docID=10929597 [Accessed December 14, 2015].
Thomas and Ayers 2014 Thomas, W.G. & Ayers, E.L., The Differences Slavery Made: A Close Analysis of Two American Communities. Available at: http://www2.vcdh.virginia.edu/AHR/ [Accessed December 12, 2014].