Functionality of the Current Implementation and Caveat, as of January 2007

Current Functionality | Limitations of Identification Procedures | Errors in the Text and Plans for Correction

Current State of the Project

The data input for the entire Encyclopédie, 17 volumes of text and 11 volumes of plates legends, has been completed. The current working version of the ARTFL Encyclopédie implements a fast and flexible metadata and full-text search engine (PhiloLogic).

What works:

Searches for articles or groups of articles by:
- Article headword
- Author
- Category of knowledge (or Classification)
- Normalized Category of knowledge (e.g., Litter. = Littérature)
- English Category of knowledge (e.g., Littérature = Literature)
- Part of speech
These searches can be run in various combinations. Ex: Author = alembert and Classification = astron
The user can also specify the type of textual object to search (in the "Type of Entry" box), including main articles (artm), sub-articles (arts), individual plate legends (plate_legend), or group of plates (plate_group). To search the entire work, leave this box empty.
The search will produce a list of articles that fulfill the search criteria. Each article headword is configured as a hypertext link to the text of the article. All available information concerning Author, Category of knowledge and Part of speech is listed after the article headword (see below, concerning limitations of the identification procedures to date).
Full text searches for:
- words
- phrases
- co-occurrences of words
- word or phrase frequency per article
The user can search the entire Encyclopédie, or search within a limited corpus as defined using the article search criteria. Ex: author = alembert, search articles for géométrique, (also try ge/ome/trique and gEomEtrique).
Full text searches yield lists of passages from the Encyclopédie with the targeted word or phrase in bold type. The result is given either as a "Concordance Report" (a passage of roughly five lines of text, currently the default) or as a "KWIC Report" (one line of text). For better contextualization, the user can then link to the paragraph, sub-article, article, or page in which the passage appears.
For co-occurrence searches, it is possible to specify whether the two terms should be located in the same sentence, paragraph, sub-article, or article.
"Frequency by Title" lists the number of occurrences for the search word or phrase in each article.
Cross-references from one article to another.
Cross-references to other articles are configured as hypertext links. Clicking on these links will yield a list of all articles that correspond to the cross-reference; the user will decide which article is actually designated by the cross-reference (based, for example, on information concerning category of knowledge). Since these cross-references are generated automatically, some of them may not work, 1) because of inconsistency in the precise name of the article (e.g., plural vs. singular, or accented and non-accented characters); 2) because the referenced article was never included in the Encyclopédie; 3) because of typographical errors in the cross-reference which prevent matching with an article headword; or 4) because something which is not, in fact, a cross-reference has been identified as such.
Links from plate legends to plate images
Article and full-text searches include the text of the plate legends. The text of the legends is linked to the appropriate plate images. Plate volumes are numbered volumes 18-25.
- A small "thumbnail" image of each plate is displayed at the beginning of the corresponding plate legend.
- The small image acts as a hypertext link to the large-format image of the same plate; simply click on the thumbnail.
Attention: Plate legends and plate images are linked automatically; in some cases, the numbering sequence pairs the wrong image with a particular legend. We would appreciate notification of such problems.
Also, the title for each plate group is currently tagged as a separate textual object and displays no text and no objects. Simply click on the Next article link to view the legends and accompanying thumbnail images. To see a list of the major plate groups, search for "Type of Entry" = plate_editorial; for a more detailed listing, try "Type of Entry" = plate_group.
Links to digitized images of each page of the text
These images are accessed by clicking the [Image] link next to the page number that appears at the top of individual pages of text. Article searches currently do not include links to individual pages; this function is only in operation for Concordance or KWIC reports.
For technical reasons, we are temporarily using page and plate images from the reprint produced by Pergamon Press, 1969, and not from the IDC microfiche copy using the for actual keyboarding of the ARTFL Encyclopédie. We have compared the text and page images and found no differences, but we cannot guarantee that the images correspond exactly to the edition used for the data capture, although, to the best of our knowledge, we believe that to be the case as suggested in our general comparison. We would be very interested in any reports of copy discrepencies between the text data and page images.

What does not work:

References to the plates from the main text
These links will be established primarily by hand at a later date. This will be one of the last steps in the project.

Editorial Concerns

Limitations in the Identification of Textual Elements

We know that there are many problems--unidentified or misidentified articles, missing author attributions, incomplete information about grammatical and knowledge categories, malfunctioning cross-references--but there is no way for our small crew to find and correct them all in a short time.
It may be helpful for users to know that, up to this point, all identifications--articles, authors, cross-references, etc.-- have been made using automated procedures based on typographical patterns. For example, large upper case letters indicate articles, small upper case letters indicate sub-articles, a single upper-case letter in parentheses at the end of a paragraph furnishes the authorship information; cross-references require somewhat more complicated patterns, but the idea is the same. (We have made only two exceptions to strict automation: we have done some hand editing 1) to insure coordination of plate images and plate legends and 2) to label the editorial texts.) Our reliance on these patterns is absolutely necessary since the Encyclopédie is such a large work. At the same time, these procedures mean that irregularities in the patterns we have used cause problems.
It should be clear from this description that some elements of the Encyclopédie will escape automated detection. For example, author attributions that are not indicated by the special author codes are less easy to detect. We have attempted to identify many of the indications such as, "This Article is by X", for example, but some may have escaped our efforts. In many cases, we will have to insert information by hand--a step that will come much later in our development process.
Please note that we have not yet attempted to integrate the results of numerous studies of the authors of articles found in the Encyclopédie or authorial attributions not found in the Encyclopédie itself. The most important of these studies include Richard N. Schwab, Walter E. Rex and John Lough, Inventory of Diderot's Encyclopédie in Studies on Voltaire and the Eighteenth Century (volumes 80, 83, 85, 91, 92, 93, 223; 1971-1984); Frank Kafker, The Encyclopedists as Individuals" in SVEC, vol. 257, 1988, and 1. "Liste des articles de l'Encyclopédie portant l'astérisque" (pp. 133 - 206) 2. "Liste des articles non signés qu'on peut attribuer avec certitude à Diderot et qui sont reproduits dans cette édition" (pp. 207 - 210) 3. "Liste des articles non signés qu'on pourrait peut-être attribuer à Diderot, mais qui ne sont pas reproduits dans cette édition" (pp. 211-219) in John Lough and Jacques Proust, Oeuvres Complètes de Didérot, Edition critique et annotée, volume 5 (Paris, Hermann, 1976). We strongly recommend that our users consult these works to verify results concerning attribution and authorship obtained from our site.

Errors in the Text

We are aware that many typographical errors have been introduced into the text during our data capture procedures. Unfortunately, due to the large dimensions of the Encyclopédie and its great semantic diversity, it is impossible to correct these errors by any normal spell-checking procedure. We have begun an Encyclopédie Corrections Project, but this is a process that will continue for some time.
Therefore, users of the ARTFL Encyclopédie will probably detect errors in the text. Occasionally, of course, this will cause difficulties in text searches, and readers will not find passages with which they may be familiar. In addition, the text includes a certain number of tags <?> which designate a missing character or symbol.
We apologize for these lapses in the text, but we have provided digitized images of the original text in order to permit readers to check the machine-readable version. When questions arise, users should consult these images, by first obtaining the individual page in machine-readable form and then clicking on the page and volume identifier at the top of the page.

Errata

We have determined that pages 364 and 365 of volume 1, containing part of the article AMMONIAC [Le sel Ammoniac, Ammoniaque (Gomme)] AMMONITES, AMNIOMANTIE, AMNIOS, AMNISIADES, AMNISTIE, AMODIATEUR, AMODIATION, AMODIER, AMOGABARE, AMOISE, AMOL, AMOLETTES, AMOME are missing.