|
|
| Press |
|
 |
 |
 |
Paris, June 16, 2005
INRIA, INSA Rennes and the Yvelines
Department Council are presenting
a new computer system at the Yvelines
Archives that revolutionizes
access to digitalized handwritten documents. The system allows
for automatic access, makes documents more available to the public
and establishes a database to store
the
work
of each user. The same application will soon be deployed at the
National Archives Historical Center and in the city of Lannion.
INRIA has long been interested in the digital document topic. The Institute has been investing for a few years in the area of digital archives, in partnership with regional institutions (department archives). The scientific competences closest to this field were found at INRIA Rennes, within the IMADOC team of IRISA and at INSA Rennes.
A new computer system was installed for the first time in the Department of the Yvelines Archives that are capable of automatically selecting, without any prior indexing by hand, handwritten documents that are of interest to a user. The new tool combines high performance handwritten shape recognition with structure or document field detection capabilities, and produces automatic annotations for each type of archival document. Using such annotations that extract multiple informations from the document's digital image, users automatically find pages of interest by family name.
When a reader asks the system to display the documents that contain the name of a person, the computer reads the entirety of digitalized handwritten documents and presents those that are closest to the request. The reader then needs only to select the desired name and the whole corresponding page is displayed. Access to the right document is completely automatic in a matter of mere seconds. The application currently concerns 19th and 20th century military number registers. Research is ongoing for automatic access by name and first name to nominative population census lists of the 19th and 20th centuries, and to allow this type of access to other, harder to process, documents.
Certain documents can't be viewed because they contain information pertaining to medical or legal confidentiality. Such is the case of military number registers less than 130 years old. These documents contain other information that are very useful to historians and archivists genealogists. The new system developed by INRIA and INSA Rennes researchers makes access to such data possible due to a virtual subdivision of the pages. The system actually masks the areas protected by law, so that everything that is in the public domain can be read. The virtual subdivision system thus provides access to a previously base of information.
It is technically impossible to achieve a complete, automatic annotation
system since all digitalized handwritten documents cannot be entirely
commented. For example, certain documents are not sufficiently well
organized for their organization to be automatically recognized,
in other documents, the handwriting may be too degraded to be legible.
Thus, the new system makes it possible to exploit the research carried
out by all users in order to circumvent such information losses.
When a reader accesses a document in the archives or over the Internet,
he or she can complement the automatic annotations with manual annotations,
known as collective annotations, using the new system. Such collective
annotations are managed on a platform and can be used in networked
work. Each reader may thus progressively record information deemed
to be significant. Using this information, other readers may be able
to automatically select documents that are then linked. For example,
a user may add the mother's family name on a birth certificate in
parish
and registry office registers. This addition thus makes this birth
certificate accessible to all persons interested in the mother's
family name. Such supplementary information can be invaluable in
genealogy research but also in history (people's signature on the documents
can reveal the literacy rate).
The collective annotation platform set up in the
Yvelines Department Archives concerns parish and registry office
registers (1,450,000 images) and military number registers (430,000
images). It will also be installed at the National Archives Historical
Center for naturalization decrees (150,000 images). The platform
may be used on all types of digitalized documents, such as nominative
lists of population census, city council meeting minutes, and so
on.
Each future reading station consists of a graphics tablet with visual
feedback, that is to say a flat screen lying on the desk like a writing
pad. The interaction between the machine and the reader is performed
via a pen interface. Even if an increasingly wide public is capable
of proficiently using a computer keyboard, the use of a pen is nonetheless
a lot more intuitive. The pen interface makes it possible to jot
down annotations, select areas and display documents without resorting
to a keyboard and a mouse. Its use corresponds to paper document
reading habits and makes it possible to easily mix paper and electronic
document consultations.
The application developed by the IMADOC team for the Yvelines Archives makes
it possible to benefit from all the capabilities of the digital pen. Tablet
PCs are well known to the general public. They are usually equipped with software
in which digital pen integration only works for office applications, and not
for image browsing software. Two reading stations using the new technology
are installed in the reading room. They provide an intimation of what the working
tools offered to Archives readers could be in the near future.
INRIA, CNRS, the University of Rennes 1 and INSA Rennes are partners
in a joint research unit called IRISA: the Institute for Research
in Computing and Random Systems. More than 450 persons, including
close to 200 scientists, are working on its premises located on one
of the Rennes university campuses. IRISA is strongly involved in
collaborations with research centers and was able to establish close
links with regional, national and international economic circles.
For more information: http://www.irisa.fr
In October 2003, the Department of the Yvelines Archives left the prestigious but not well adapted premises of the Grande Ecurie du Roy in Versailles, and moved to a new building at Saint-Quentin-en-Yvelines. In addition to an increased storage capacity, the new building emphasizes document accessibility via new technologies: digitalization of the most consulted documents (2 million files), free consultation in the Department Archive reading room equipped with 54 computers or over the Internet (www.archives.yvelines.fr). In addition, owing to a partnership with INRIA and INSA, the Yvelines Department Archives puts innovative search tools at the disposal of its users (collective annotation platform, handwriting recognition, etc.) that facilitate access to digitalized archives.
Version PDF of the Press Release
| IRISA Bertrand Coüasnon Tel.: +33 2 99 84 74 11 |
INRIA Laurence Hermant Stratéus for INRIA Tel.: +33 1 40 41 56 11 Vincent Coronini Tel.: +33 1 39 63 57 29 |
| INSA Claire Antony Tel.: +33 2 23 23 86 29 |
Conseil général des Yvelines / Archives
des Yvelines Laurence Volf Tel.: +33 1 39 07 71 53 |
