This prototype of the first volume of the Encyclopédie is presented by the ARTFL Project (the Project for American and French Research on the Treasury of the French Language), at the University of Chicago.
PLEASE NOTE that this prototype does not include all of the functions that will be part of the final version, all formatting and identification to date has been completed by automated procedures, and the text has not yet been proof-read and corrected. This means that a certain number of errors will occur, both in the text itself and in functioning of the database. We feel, nevertheless, that the prototype provides a fair idea of some of the new possibilities that the on-line environment creates for readers and scholars of the Encyclopédie.
Operation of the ARTFL Encyclopédie is largely intuitive. Therefore, users who are familiar with the Web may wish to proceed directly to the Search Form. To consult documentation describing the ARTFL Encyclopédie, either scroll through the following text or select individual topics from the list of headings immediately below.
All users may wish to consult the explanation of the two types of Images, that are a part of the ARTFL Encyclopédie, the facsilimile page images and the plate images (planches).
Send comments and questions to: mark@gide.uchicago.edu
The Search Form allows users to define and submit queries using several different parameters. The categories Headword, Author, Classification and Part of Speech may be used singly or in combination with one another. If no restriction is desired for a particular category, the box should be left blank.
By defining searches in different ways, users can limit the results of their queries to manageable proportions and obtain results more quickly. It should be noted that the memory capacity on some work stations is not sufficient to permit full display of all records from very broad queries.
A search for "Author: d'alembert" (or "alembert", or "Alembert" or "d'Alembert"), for example, yields 200 records. This may, in some cases, be a useful result. To obtain more focused results, however, the user may want to pair "Author: d'alembert" with "Classification: astro". This query yields 21 records, not only a smaller amount of text to read, but also a more coherent group of articles.
Accents and upper case letters.
Accents should not be used. The "Search Form" will not accept accents. Entering accents will result in a failure of the search. Therefore, although the headword "APPUYÉ" contains an accent, the correct search query will be "appuye".
Furthermore, the "Search Form" is not case sensitive; there is no distinction between upper- and lower-case characters.
Search results will appear as a list of all headwords for entries that correspond to the specifications of the search. Main articles appear in all upper-case characters. Sub-articles appear in lower-case characters. The user may click on individual headwords to see the text of the entries listed.
Note: clicking on the headword of the main article will also yield the text of the accompanying sub-articles.
The use of the "logical operators" and of "pattern matching" will allow users to define searches with more precision. One note of warning: most articles have been identified only by the initial word. In most cases, it will not be possible to search for subarticles that are qualified by classification or in other ways. Thus "Abaissement du Pole" is listed only as "Abaissement"; "ABARIME ou ABARIMON" only as "abarime".
Logical Operators.
It is possible to define searches with greater precision by using the logical operators "AND" and "OR" (in upper-case characters).
To obtain articles concerning only ancient geography, for example (and excluding modern geography), "Classification" might be defined as "geog AND anc", which yields 128 records, rather than 609 for "geog" without the "anc" qualifier.
To obtain articles in related fields, the "OR" operator is useful. Thus "Classification: chasse OR fauc" would give all articles related to falconry and to hunting more generally, a pairing suggested by the article "ACHARNER, v. act. (Chasse & Fauc.)". The results to the query would combine all records from separate searches for "chasse" and "fauc".
Pattern Matching.
Pattern matching can also help define searches. By default, a search in a field will match all substrings. Thus, searching for "action" will find "action", "actionaire" and words containing "-action" such as "abstraction" and "attraction".
To anchor searches to the beginning of a word, use the "^". Thus, searching for "^action" will find "action" and "actionaire". To anchor the search at the end of the word, use the "$". To anchor both the beginning and the end, enter "^action$", which will find only "action".
These specifications are repeated in an explanation just below the Field Search Form.
The "Search Form" includes several lists that may be useful for formulating queries. These lists have all been generated automatically and have not been edited in any way. They contain errors and are incomplete.
Users can click on "Headword" to obtain a list of all main articles identified in volume one. Users can also click on "Classification" to obtain a tabulation of the categories that have been identified for all articles. This list may be useful for identification of common categories and their abbreviations.
Below the "Search Form", users will find lists of "Author Names with Frequencies" and "Parts of Speech". Again, these lists have been generated automatically and are incomplete.
The first-volume prototype includes two types of images: images of the plates that are cited in articles in the first volume, and facsimile page images for each page of the first volume. The mode of consultation for each type of image is described below.
Imaging is incomplete at the time of the public release of the prototype. In the case that an image is not available, a default image will appear. In the case of the page images (about 700 pages available), the default is the title page of volume one. For the plate images (about 15% complete), the default is one of the "Géographie" plates. We will continue to make additional images available as scanning is completed.
Note: Image display--format, resolution, and speed--will vary in accordance with the capabilities of the work station being used and the network connection. Images are formatted in jpeg; page images, when compressed, run about 550K and plate images about 400K.
When consulting an article, the user will see the page number indicated in parentheses after the article headword. Clicking on this page number will call up the machine-readable text for that page. To obtain the facsimile page image, the user then clicks on the underlined page number at the top of the screen.
These page images are particularly important because they enable users to verify the machine-readable form against the original text. Some elements of the text must be omitted during the data-capture process. These include complex tables and mathematical formulae, musical scores, and unusual print characters. Many of these elements have been marked with a tag, <?> or <omit> in the machine-readable text.
For the purposes of the prototype, plate images will be accessible only through references in the articles. At a later stage in development, we will install a mechanism to permit users to browse through the images by subject.
In the articles, plate references have been identified and underlined automatically. The links to the appropriate plates have been established by hand, based on the information in Schwab's Inventory of Diderot's Encyclopédie, volume VII. The user should click on the underlined plate reference to obtain a small reduction of the corresponding plate, followed, in some cases, by a portion of the text of the plate legend. The user may click on the small image to obtain a much larger image of the full plate. For even greater magnification of details, the user can use a "crop and zoom" procedure.
The prototype represents a first stage of development for the ARTFL Encyclopédie. Additional functions will be added at a later date, and there will be an extensive editing procedure. Thus, several features that will be part of the final version are currently unavailable.
Users should recognize that all processing of the text for the prototype has been completed through automated procedures with no hand editing. This produces several types of errors that will be corrected at a later date using automated detection and either manual or automated correction procedures.
All identification of articles headwords, article cross-references, and plate references has also been accomplished through automated procedures. The results of these procedures have generally been excellent, resulting in correct identification of nearly 95% of the article headwords and approximately 80% of the article and plate references and author attributions. The success rate for "Classification" and "part of speech" is currently somewhat lower. We therefore acknowledge that omissions and erroneous identifications have occurred; they will be corrected at a later date.
Articles resulting from the contributions of several authors pose a particular difficulty. Although such "entries " are listed in Schawab's Inventory, there is no way of identifying them automatically. These entries within articles will remain unidentified until we undertake hand editing procedures.
We have not yet completed imaging procedures for the first volume. Requests for certain images will thus yield a default image (either a default plate image or, in the case of text pages, the title page). Also, the plate images are not fully navigable. At a future date, users will have the possibility of searching for images by category. At present, the only access to images is through the individual articles which contain references to them.
The prototype does not currently allow users to conduct full-text searches for words or combinations of words. This is one of the principle features of the main ARTFL database, and it will be included in later stages of development of the Encyclopédie. Eventually, users will be able to define a group of articles (using the procedures that are currently in operation) and to conduct searches within this corpus.
For the purposes of the prototype, we have entered only a small portion of selected plate legends. This text, when available, appears with the reduced plate image. In future stages of development, the legends will be integrated as a part of the searchable text. They will also be displayed with the corresponding plate images, as is presently the case.