TOWARD A SEARCH ENGINE OF MULTIMEDIA

PRESENTATIONS

Belkhir Abdelkader, Bouyakoub Fayçal M’hamed, Bouyakoub Samia

Faculté d’Electronique et d’Informatique,USTHB, B.P. 32 El Alia, Alger, Algérie

Keywords: SMIL, multimedia presentations, meta data, indexation, Dublin core, RDF.

Abstract: Facing to the multi-media explosion, it is necessary to develop new information access methods which

allow both fast and natural information access. To take into account this aspect for multimedia

presentations, it needs to implement a graphical interface facilitating the cross, the search and the

visualization of the medias. Moreover, this interface must answer the readers aspirations for both

conviviality and easy use. This article presents an interface paper for multimedia presentations and proposes

an indexing method for exploration on the WEB.

1 INTRODUCTION

The research of video is a difficult task because it is

essentially about a time-dependent media. At one

instant 't', its visualization gives access to a part of

information, and doesn't permit to deduct the global

context. On the other hand, the sequential searching

on a video proves to be expensive in terms of time

(the length of the sequence can go until several

hours) and in resources (CPU time, bandwidth…). It

is therefore necessary to offer mechanisms

permitting a fast and efficient access to the

information contained in a video, without having to

cover it entirely.

Therefore, the implementation of a diffusion

video environment must take into account the

following points: the primitives of description of the

video, the organizational methods and the

implementation of a convivial interface.

Our solution, called "paper interface" was

developed in this way. It is a Web interface whose

appearance is inspired strongly from the principle of

the classic newspaper. A “paper interface” is a Web

page permitting to organize the videos in several

themes. The page associated to a given theme

contains, for every video, an image capture of this

video as well as the first lines of an associated

explanatory text. The organization and the page

setting of the interface are inspired from the

principle of the classic newspaper (setting in

columns, choice of the police of the text and

titles…), from where the appellation of paper

interface. This resemblance to the newspaper confers

to the interface a familiar and convivial aspect.

Section 2 presents our solution and its

implementation. In order to complete this solution,

section 3 provides a draft for indexing multimedia

presentations. We finish by a conclusion in section

2 THE PAPER INTERFACE

The paper interface comprises a main Web page

permitting to present the different themes, and

containing some links toward the Web pages

associated to every theme. To every theme is

associated a folder containing the recordings video,

the associated explanatory texts, as well as the set of

the files necessary to the good performance of the

interface.

The choice of a Web solution is motivated by the

evolution of Internet and the success of the World

Wide Web making it a universal platform for

multimedia diffusion with least cost. On the other

hand, one of the advantages of the Web is its

capacity to separate the presentation of the

information from its content, in order to better

secure the system.

The proposed solution uses the HTML language,

jointly to ActiveX technology permitting to integrate

the ActiveX of RealOne Player to the paper

interface.

487

Abdelkader B., Fayçal M’hamed B. and Samia B. (2006).

TOWARD A SEARCH ENGINE OF MULTIMEDIA PRESENTATIONS.

In Proceedings of WEBIST 2006 - Second International Conference on Web Information Systems and Technologies - Internet Technology / Web

Interface and Applications, pages 487-490

DOI: 10.5220/0001251904870490

 SciTePress

A language of scripts permits to manipulate the

methods of the ActiveX. Our choice was about the

language Java Script because it permits to

manipulate the methods of the ActiveX of RealOne

Player, and assure a portability (inherited of the Java

language) on all platforms.

2.1 Why SMIL Language?

The format of specification used by the authoring

environment has an important role in the

implementation of the system, since the popularity

of the chosen language affects the possibilities of

diffusion of these presentations on a large scale,

notably on the Web, that’s what justifies the choice

of the HTML language. However, the absence of

synchronization mechanisms makes that HTML is

inappropriate to the modelling of synchronized

multimedia presentations where the temporal

component is predominant.

The interface paper integrates several

synchronized multimedia flows; it is obviously this

aspect of synchronization that justifies the choice of

the SMIL language (W3C recommendation) (W3C,

1998) as specification language within our

application.

2.2 Implementation of the Paper

Interface

The principle of the paper interface consists in

presenting a description of the video in order to give

a preview on its semantic content. The chosen

primitives of description are a segment of the

explanatory text associated to the video and a picture

extracted from this video sequence.

The extraction of the text consists in extracting

automatically the first lines of the text associated to

the video, and to integrate them to the paper

interface.

However, we don't have a mechanism to capture

automatically the video picture. The proposed

solution is to create this picture artificially, using the

SMIL language and ActiveX technology. The zone

of the picture is in fact the display window of the

ActiveX integrated to the Web page. The principle

consists in starting, for every video, a specific SMIL

presentation (noted VideoCap) permitting to play the

video during one sufficiently short time interval (of

the order of 0.2 to 0.7 seconds) to capture the first

picture of it. We will keep this picture displayed

thereafter on the screen, what permits to get a frozen

picture. The VideoCap presentation is a SMIL

presentation generated automatically at the level of

the interface according to a preset model that will be

given thereafter.

These operations requires the automatic

generation of HTML code, jointly to Java Script

commands, and the integration of this code in the

Web page. The internal data structure of a page

associated to a given theme is a dynamic chart

defined in HTML. The internal organization of the

interface is illustrated by Figure 2.

The general mechanism of working is illustrated

by the following sequence:

- The HTML document representing the paper

interface contains an ActiveX control of

RealOne Player permitting to launch

automatically a specific SMIL presentation

permitting to capture a picture of the video,

noted "VideoCap.smi".

- The SMIL document "VideoCap.smi" permits

to play the video sequence "video.rm" during a

very short time (here 0.7s), and to keep the last

picture displayed on the screen thanks to the

attribute fill = "freeze." It contains a link

(element area) toward a Ram file "Lien.ram."

- The file "Lien.ram" is a RAM file whose first

line makes reference to the main SMIL

presentation permitting to play the selected

video.

- The main SMIL presentation

"Presentation.smi" permits to play the selected

video. It contains a link (element area) toward

an external HTML page (informations.html)

containing informations concerning the video.

The files necessary to the interface working are

generated automatically by a software module "the

interface generator".

Figure 1: An example of the paper interface.

WEBIST 2006 - WEB INTERFACES AND APPLICATIONS

488

The interface generator is a software tool

permitting, from a video and an associated

explanatory text, to generate a new entry

automatically in the interface paper.

This module allows the creation of themes and

the addition of new presentations to an existing

theme. The solution rests on mechanisms of

automatic generation of HTML or SMIL code.

The implementation has been achieved in a

SMIL presentations diffusion system on demand,

where the customer can have a restricted right of

consultation, or have in more, a right of creation,

addition and deletion of presentations in the

interface paper, according to the type of terminal

used and to his statute. The spreading of our

interface paper in client/server architecture is based

on the requests exchange and SIP answers (Handley,

1999), which contain some SDP messages (Handley,

1998) that transport the useful information.

The integration of the paper interface to a

wireless video diffusion platform permitted to get a

complete video diffusion system on demand

answering the needs of the authors (insertion of new

presentations) and of the readers (paper interface

facilitating research and the visualization of the

medias). The layout and the choice of the fonts,

inspired from the classic newspaper, permit a good

legibility of the text, in order to adapt it to the

reduced display surface of the PDA.

However, in order to allow the users to reach

these presentations via the Web, it is necessary to

propose indexing mechanisms of these presentations

on the Web, and to permit their reference by

searching engines.

We will propose an approach of indexing of the

SMIL presentations within the interface paper.

3 INDEXING THE PAPER

INTERFACE PRESENTATIONS

Since some years, the progress in data processing

made that the Web became an immense field of

knowledge (Sénac, 1996). The important volume of

the numeric data and the diversity of their formats

makes that it becomes more and more difficult to

find information on the Web, even when this one

exists. In general, specialized software "the

searching engines", permit to explore the Web and

to automate the process of research. In order to make

the numeric data accessible to the users, it is

necessary to have information on the documents and

their semantic content. In the case of multimedia

documents, these information become indispensable;

otherwise these medias would be inexploitable by

the search engines (Hunter, 1998) (Hunter, 2001).

The solution proposed by the W3C is to use the

Meta data to describe the data contained in the Web.

The Meta data are used by the systems of content

management (CMS: Content Management Systems)

to publish, manage, search for, reuse, distribute and

to publish multiple contents (texts, pictures, video,

etc.) (Hunter, 2001).

The research made in this sense led to the RDF

recommendation of the W3C (Lassila, 1999). RDF

(Resource Description Framework) is a tool to

encode, exchange and reuse of the Meta data

structure. It provides the inter-operability between

the applications that exchange non comprehensible

information by the machines on the Web. It

increases the easiness of automatic treatment of Web

resources (Lassila, 1999) (Hunter, 1998). However,

RDF has some limitations: In the one hand, it’s

cannot refer to a portion of a media object; In the

other hand, the Meta data are defined in an external

file to the resource, whereas it would be preferable

to have some information at the level of the

resource, as in the case of the SMIL language.

3.1 Integration of the Meta Data in

the SMIL Documents

One of the major contributions of SMIL is the

possibility to integrate some Meta data within the

SMIL document.

We have seen that RDF didn't permit to reference

portions of continuous medias. The SMIL language

permits to get round this problem using the links of

navigation (the “anchor” element of SMIL 1.0 and

“area” of SMIL 2.0) with attributes “begin” and

“end” permitting to delimit the portion of the wanted

media. The use of these elements permits to assign a

Figure 2: Organisational structure of the paper interface.

TOWARD A SEARCH ENGINE OF MULTIMEDIA PRESENTATIONS

489

reference toward a segment of presentation while

using the attribute “id” to identify the segment, and

the attribute “href” for referencing the segment.

SMIL permits in addition to refer to a temporal

part of the presentation, this part can be either a

simple media element (audio, video, text), or

composite (element “par” or “seq”), by associating it

to a unique identifier.

SMIL permits to reference the multimedia

content in an efficient manner, by permitting to

segment a media in several portions and associating

a description to each of them. However, the list of

attributes of the “meta” element stills open, therefore

this information cannot be exploited efficiently by

any search engine. The ideal would be to be able to

describe these informations in a standard manner

independently of the application domain.

We opted for a hybrid approach based on the use

of the attributes defined by the Dublin Core

Metadata Initiative (or rather one subset of these

attributes), in order to describe the Meta data in a

SMIL 2.0 document. The Dublin core is a set of

Meta data attributes defined in order to permits the

exchange of Meta data and the interoperability

between several domains of application.

The interface generator translates the

informations given by the author to Meta data, and

inserts them in the SMIL document thanks to the use

of the module of Meta data of SMIL 2.0. We defined

a model of Meta data containing a subset of the

elements of Dublin core.

Every Meta data will be translated to a meta

element, where the name of the property is an

element of the Dublin Core, preceded by "DC: " in

order to signal that it is about an element of the

Dublin Core (Powell, 2001).

The SMIL language being drifted from XML, the

integration of the elements of the Dublin Core

passes by the integration of the corresponding name

space.

The elements of the Dublin Core permit to define

the Metas data associated to any domain of

application. These elements are both comprehensible

by the machines and by the users. However, in order

to assure a maximal compatibility, the Dublin Core

should integrate sets of predefined values associated

to every element. Some research are in progress at

the level of the DC (Hillmann, 2003) in order to

solve this problem.

4 CONCLUSION

The originality of our solution resides in its

simplicity of use. Indeed, thanks to a good

organization and a suitable layout, the user's

attention is carried quickly on the picture

representing the video; thereafter, the reading of

some lines of the associated text can help him to

decide, while giving him a global view of the

content. The use of this interface is intuitive, and

doesn't require any particular knowledge. Besides,

the proposed solution contains an automatic

mechanism permitting the generation and the

automatic update of the interface paper to a software

tool: the interface generator.

We also proposed a new approach of description of

the multimedia resources, based on the use of the

elements of the Dublin Core jointly to the SMIL

language. This solution, applied to the interface

paper, permits the indexing of the presentations and

their reference by the search engines, notably on the

Web.

REFERENCES

Handley, M., et Jacobson, V., 1998. SDP: Session

Description Protocol. RFC 2327, IETF.

Handley, M., Schulzrinne, H., Schooler, E., Rosenberg, J.,

1999. SIP: session Initiation Protocol. RFC 2543,

IETF.

Hillmann, D., 2003. Using Dublin Core. DCMI

Recommended Resource, on line at

http://dublincore.org/documents/2003/08/26/usageguid

e/.

Hunter, J., Iannella, R., 1998. The Application of

Metadata Standards to Video Indexing, Second

European Conference on Research and Advanced

Technology for Digital Libraries, Greece.

Hunter, J., Little, S., 2001. Building and Indexing a

Distributed Multimedia Presentation Archive using

SMIL. Proceedings of the 5th European Conference on

Research and Advanced Technology for Digital

Libraries.

Lassila, O., Ralph, R., 1999. Resource Description

Framework (RDF) Model and Syntax Specification.

W3C Recommendation, on line at

http://www.w3.org/TR/1999/REC-rdf-syntax-

19990222.

Powell, A., Wagner, H., 2001. Namespace Policy for the

Dublin Core Metadata Initiative (DCMI). On line at:

http://dublincore.org/documents/2001/10/26/dcmi-

namespace/.

Sénac, P., 1996. Modelling logical and temporal

synchronization in Hypermedia Systems. IEEE journal

on selected areas in communications.

W3C recommendation, 1998. Synchronized Multimedia

Integration Language (SMIL) 1.0. on line

at:http://www.w3.org/TR/REC-smil.

WEBIST 2006 - WEB INTERFACES AND APPLICATIONS

490