IMPLEMENTING MULTILINGUAL INFORMATION

FRAMEWORK IN APPLICATIONS USING TEXTUAL DISPLAY

Satyendra Kumar Gupta, Samuel Cruz-Lara, Laurent Romary

Loria (UMR 7503) / INRIA Lorraine, Campus Scientifique – BP239, 54506 Vandoeuvre-les-Nancy, France

Keywords: MLIF, Interactive multim

edia applications, Natural language display, Localization

Abstract: This paper presents implementation of MLIF [1] (Multilingual Information Framework), a high level model

for describing multilingual data across wide range of possible applications in translation/localization process

within several multimedia domains (e.g. broadcasting of interactive multimedia applications), natural

language interfaces, geographical information systems for multilingual communities.

1 INTRODUCTION

Linguistic information plays an essential role in the

management of multimedia information, as it bears

most of the descriptive content associated with more

visual information. Depending on the context, it may

be seen as the primary content (text illustrated by

pictures or videos), as documentary content for

multimedia information, or as one among several

possible information components in specific

contexts such as interactive multimedia applications.

Linguistic information can appear in various

rmats: spoken data in an audio or video sequence,

implicit data appearing on an image (caption, tags,

etc.) or textual information that may be further

presented to the user graphically or via a text to

speech processor.

In this context, dealing with multilingual

formation is crucial to adapting the content to

specific user targets. It requires one to consider

potential situations where the linguistic information

contained in a multimedia sequence is either already

conceived in such way that it can be adapted on the

fly to the linguistic needs of user, or by using an

additional process where content should be adapted

before presenting it to the user.

Finally, there are a wide variety of applications

within

which multilingual information may appear,

which supports development and implementation of

generic framework, MLIF, for dealing with

multilingual content: subtitling of video content,

dialogue prompts, menus in interactive TV,

descriptive information for multimedia scenes,

karaoke management, etc. Such information should

be considered in the light of the experience of more

specialized communities traditionally dealing with

multilingual content, namely the translation and

localization industry.

2 MULTILINGUAL

INFORMATION FRAMEWORK

The Multi Lingual Information Framework (MLIF)

is designed with the objective of providing a

common platform for all the existing tools

developed by the different groups (LISA [2], OASIS

[3], W3C [4], ISO [5]). It promotes the use of a

common framework for the future development of

several different formats: TBX, TMX, XLIFF,

Timed Text, TMF, etc. It does not create a complete

new format from scratch, but suggests that the

overlapping issues should be handled independently

and separately. It will save time and energy for

different groups and will provide synergy to work in

collaboration. Presently, all the groups are working

independently and do not have any mechanism for

taking advantage of each other’s tools. MLIF

proposes to concentrate on only those specific issues

that are different from others and specific to one

format only, so it will create a smaller domain for

the groups’ developers. It gives more time to

concentrate on a subset of the problems they are

currently dealing with and creates a niche that helps

in providing a better solution for problems of

multilingual data handling and translation issues.

172

Kumar Gupta S., Cruz-Lara S. and Romary L. (2005).

IMPLEMENTING MULTILINGUAL INFORMATION FRAMEWORK IN APPLICATIONS USING TEXTUAL DISPLAY.

In Proceedings of the Seventh International Conference on Enterprise Information Systems, pages 172-175

DOI: 10.5220/0002544301720175

 SciTePress

MLIF deals with the issues of overlap between

the existing formats. MLIF involves the

development of an API through which all these

formats are integrated into the core MLIF structure.

This is done through the identification and a

selection of data categories [10] (DCs) as stated in

ISO DIS 12620-11 (in ISO/TC 37/SC 32). MLIF can

be considered as a parent for all the formats that we

have mentioned before. Since all these formats deal

with multilingual data expressed in the form of

segments or text units they can all be stored,

manipulated and translated in a similar manner. This

kind of data can easily be stored in data categories

and in terminological mark-up. The results of IST

SALT project [6] clearly show that it is not difficult

to edit, store and reuse data categories. The SALT

project combines two interchange formats: OLIF [7],

which focuses on the interchange of data among

lexbase resources from various machine translation

systems, and MARTIF [8], which facilitates the

interchange of termbase resources with conceptual

data models ranging from the simple to the

sophisticated. It provides a graphical user interface

that can be used to access or to define new data

categories or modify them.

3 IMPLEMENTATION OF MLIF

Some multilingual content stuff has been integrated

in the demonstrations developed. As all the Data

Categories related to Digital Media has not been yet

identified and defined, in both applications

multilingual content has been encoded in XLIFF

(which is completely interchangeable to MLIF with

the help of XSLT styelsheets, we have written).

3.1 XMT-O Location Round Trip

Using MLIF

In this application, we have obtained by means of

translation/localization process, a document in

English (Figure 3) corresponding to an XMT

document that is in French (Figure 2). Figure 1

shows how the localization process is performed.

Figure 1: XMT Localisation Round Trip

1. The original XMT

French

document containts

linguistic information in French.

2. Transformation of XMT

french

document

into MLIF

french

document.

3. Transformation of the MLIF

french

document into an XLIFF

french

document.

4. By using existing XLIFF environment, a

professional translator performs French-

English translation. We obtain XLIFF

english

document.

5. /6. Transformation of XLIFF

english

document into an MLIF

english

document.

7. /8. Transformation of the MLIF

english

document into XMT

english

document

Figure 2: An Interactive MPEG-4 Presentation in French.

IMPLEMENTING MULTILINGUAL INFORMATION FRAMEWORK IN APPLICATIONS USING TEXTUAL

DISPLAY

173

Figure 3: Interactive Presentation in English.

3.2 Identifying Monolingual Content

in XMT-O

Identifying monolingual content in XMT-O

document may be considered from two points of

view:

1. Textual information related to metadata,

2. Textual information related to data (i.e.

subtitles).

For simplicity, we will consider only textual

information related to data, that is, textual

information that may be associated under the form

of subtitles, to a multimedia presentation.

Thorough study of XMT-O reveals that all textual

information related to data (i.e. subtitles) in XMT-O

document, is included in the “textLines” attributes of

the <string> tag, for example:

<string dur=”800s”

…

textLines=”" Presentation of MLIF;;”

…/>

So, it is rather easy, by parsing a XMT-O

document, to retrieve all monolingual textual

information related to data. Developing a XSLT

stylesheet, transforming a XMT-O document into

MLIF document or vice-versa is easy task. However,

we must verify that

• The XSLT stylesheet preserves the

original XMT-O structure when

transforming into a MLIF document,

• The XSLT stylesheet takes into account

all Data Categories related to XMT-O

original document.

Though the task of identifying monolingual

content inside an XMT-O document is rather easy,

identifying Data Categories is a complex task.

The very first step is to setup an DCS (Data

Category Specification) related to Digital Media.

This activity is very complex because we have to:

• Identify all existing DCs that may be used

in the context of Digital Media knowing

that several Data Categories may be

common to several different kinds of

language resources,

• Very few DCs related to Digital Media

have been identified and defined.

Identifying and defining DCs is complex

process because

o Digital Media experts have to be

involved in identification of DC

for multimedia,

o DC experts must approve all the

DC identified (and proposed) by

Digital Media experts,

o It is an official ISO normalization

process that takes time.

3.3 Historical Information Display in

MPEG-4 Application

In the framework of ITEA project “Jules Verne”, an

MPEG application has been developed. This

application shows an interactive globe in the middle

of the screen and two other display boards on either

side of the globe. In the display board on left, it

shows the map of the country chosen and on right

side it shows the historical information about the

country chosen. This also has some flags on the

display board on right. User can click on any of the

flags and the historical information is displayed in

the chosen language. This information is encoded in

XLIFF/MLIF document (both formats are

completely interchangeable with the help of

stylesheets). This information is extracted from the

document, on run time, with the help of chosen

parameters for language identifier and country name

identifier.

Figure 4: Information Display in French.

Figure 4 shows one screenshot of MPEG-4

Multimedia presentation. In this screenshot, the

textual information (in French) appearing at the right

side of the screen has been retrieved from

ICEIS 2005 - HUMAN-COMPUTER INTERACTION

174

XLIFF/MLIF document. Figure 5 shows the same

application once user has chosen to see information

in English.

Figure 5: Information Display in English

4 USE OF XSLT TO DEAL WITH

DIFFERENT MULTILINGUAL

CONTENT FORMATS

Figure 1 has shown that dealing with multilingual

content means transforming one document (XMT

for instance in figure 1), first into MLIF document

and then into XLIFF document. The XSLT

stylesheet XMTToMLIF is a very simple way to

obtain MLIF documents (canonical from) from

XMT-O document. It should be noted that XSLT

stylesheet allows to retrieve only textual information

and original structure in not preserved.

We have written many stylesheets for the purpose

of all the transformations shown in figure 1. Besides

these stylesheets, we have also written some more

stylesheets to transform documents from/to MLIF

to/from TMX, I18N.

5 POTENTIAL

IMPLEMENTATION AREAS OF

MLIF

Textual data is the primary vehicle in which

information (for user interface on web or any digital

media) is encoded. There is pressing demand for

facilitating the access to and translating/localizing of

the linguistic data contained in these applications for

multilingual communities. Wherever we have

application serving to multilingual communities and

displaying textual information, it is inevitable that

the textual information is stored, handled and

displayed in proper format and language of user’s

choice, irrespective of medium used for displaying.

Extraction of linguistic data (while keeping

formatting, displaying and other information in

separate tags) from multiple sources and languages

(books, periodicals, newscasters, television

programs, e-learning resources, etc.) and fusion into

a user-chosen language requires understanding of

the data and easy handling.

We have identified wide range of potential

implementation of MLIF. MLIF can be used in e-

learning, interactive television programs and any

other application having user interface. It is very

helpful for future interactive television broadcasting.

It presents ample opportunity for giving value to

different languages and cultures, as is the case in

Europe and Asia.

6 CONCLUSIONS

In this paper, we have shown implementation of

MLIF with different multimedia applications to

display linguistic information. With the help of style

sheets, use of data categories while transforming

document to/from MLIF gives good results. Results

of our current work are encouraging and we are

working to use MLIF with STB (Set Top Box) for

advanced IDTV.

REFERENCES

Cruz-Lara, S., Gupta, S., Romary, L., 2004. Handling

Multilingual content in digital media: The

Multilingual Information Framework. In EWIMT-

2004, European Workshop on the Integration of

Knowledge, Semantics and Digital Media Technology.

LISA. www.lisa.org.

OASIS. www.oasis-open.org/home/index.php.

W3C. www.w3c.org.

ISO. www.iso.org.

SALT. Standards-based Access to multilingual Lexicons

and Terminologies.

http://www.loria.fr/projets/SALT/saltsite.html

Open Lexicon Interchange Format. www.olif.net

MAchine-Readable Terminology Interchange Format. ISO

12200: 1999

XMT: MPEG-4 Textual Format for Cross-Standards

Interoperability. Kim, M.; Wood, S. IBM T.J. Watson

Research.

Computer Applications in Terminology – Data Categories.

ISO 12620.

http://www.iso.org/iso/en/CatalogueDetailPage.Catal

ogueDetail?CSNUMBER=2517&ICS1=1&ICS2=20&

ICS3

IMPLEMENTING MULTILINGUAL INFORMATION FRAMEWORK IN APPLICATIONS USING TEXTUAL

DISPLAY

175