TCDL Bulletin
Fall 2007
Volume 3   Issue 3

Metin Çavus
Ph.D. candidate
YTU Faculty of Art and Design
Art and Design Programme
34349
Istanbul, Turkey
<metcavus@yahoo.com>

Abstract

The volume of information in digital video libraries (DVL) contributes to many problems experienced by users. Because of the quantity of the information, the process of searching and browsing videos has to be completed in several steps. One of these steps is preview presentation of the target video, where the video is summarized using different techniques. Several research projects have been conducted to improve video summarization, but most of the output of this research has focused on how usability is affected by information architecture. However, we believe aesthetic satisfaction of the user is also an important issue to consider for generating better video summarization. This is an issue rarely studied in the human computer interaction (HCI) field. There are a number of research questions to be answered in this regard, but the doctoral research discussed in this paper focuses specifically on the Turkish user and tries to answer the question: "Is localization a criteria for aesthetic satisfaction of user interfaces (UIs) in digital video libraries?" In the course of our research project, we will monitor a minimum of 150 novice and 50 expert Turkish users. Our work is being sponsored by Turkish Advertiser Foundation.

1. Keywords

Digital video library (DVL), user interface design, video summarization, localization, cultural usability.

2. Motivation for the Research

Digital libraries (DLs) have been a noticeable field of study for academic world in the last few years. Among the DL studies, those that focus on digital video libraries (DVLs) have presented an important challenge not only for professional librarians and scientists but also for interaction designers and usability engineers. The nature of the medium (the large volume of the files, the temporal nature of combined video, image and sound, etc.) introduces some new problems and increases the expectations [14]. While the studies of system engineers and librarians have caused efficient development in the field of DLs [13], research on user interface design and usability of DVLs has not been broad enough thus far [4], [6], [7], [11]. In addition, research to evaluate user satisfaction of a design for a specific area of the world and for people with a specific cultural background has not yet been realized in the field of DVLs. The main motivation of our research, then, is to discuss the possibility of improvement of a design system, e.g., a DVL, that considers cultural criteria.

The ability to adapt a user interface for culturally diverse communities is an issue with increasing importance. Defining the local cultural perspective has been vital for solving cultural usability problems, especially for the domain-specific interface applications with high-risk content. Therefore, we are conducting series of case studies involving the Turkish user preferences. Our initial effort is to describe the interface preferences of the Turkish user and to present a design guideline for localization.

After we discuss cultural features that can be decisive for the localization process, we will design novel DVL user interfaces. This will then be compared with the interfaces of existing DVL systems, and the validity of the results will be examined. For measuring the usability of the system, we will ask a minimum of 150 novice and 50 expert Turkish users for their subjective opinions. Efficiency of use and subjective satisfaction will be the priority attributes for evaluating our system.

3. Biographical sketch

DVLs are places where digital videos with various content and of different duration are stored. The aim of a DVL is to provide an opportunity for users to interact with a system when they try to complete different tasks. Although there are similarities between DVLs and DLs, there are very significant differences as well. The main reason for these differences is the non-textual feature of digital video [14]. This characteristic defines the most comprehensive effort to study the field of DVL. Digital video storing, retrieval, segmentation, indexing, metadata creation and feature detection are just some of the issues being discussed today [6], [11], [14], [20]. Among them the video summarization issue is also directly affected by the non-textual nature of the video medium.

Gathering video by browsing in a DVL is a process defined by different challenges and must be completed in several stages [7]. The degree to which a user can visualize returned data and evaluate search results determines how usable the system is. In this phase of the process, presenting the summary of the video is an essential feature for evaluating the content, which shortens the viewing time and increases efficiency and usefulness of the system. For creating a summary of a video, we should consider not only visual but also linguistic and audio gist.

3.1 DVLs and Basic Summarization Techniques

There are several important DVL projects and visual summarization solutions; four of these are listed below:

1. The Open Video Digital Library (OVDL) [3]: Videos in OVDL are summarized using three different techniques: 7-second experts, storyboards and Fast Forwards. A comparative study of Fast Forward speeds showed that Fast Forward speed of 64X is best for visual summarization [22].
2. The Fischlar Digital Video Library (FDVL): For visual summarization of videos in FDVL the keyframe-based method is used and eight different keyframe-based UIs have been designed [20]. Among the three highlighted approaches, the first UI is based on two different layers: an overview layer with 32 keyframes and a detailed layer with full shot layer keyframe (one keyframe from one shot). In the second highlighted UI, large-size keyframes automatically flip one by one in a slide show. In the last UI, the screen displays a small number of keyframes selected from throughout the video content, providing an overview [11].
3. The Informedia Digital Video Library [2]: Visual summarizations in this project are based on storyboards, which are keyframes arranged in chronological order. Within a storyboard, the user can inspect keyframes and display videos [4].
4. The ECHO Historical Film Archive [1]: For visual summarization of a digital video from the project, a sequence of moving images is prepared offline. The main goal of the summary is to maintain the flow of the story. The abstract is usually up to 8% of the length of the original video [17].

There is also one significant approach for enhancing the user's viewing experience [12], and in this project an enhanced browser is designed and various video browsing techniques are discussed. Time compression is just one of them, and within this technique, in order to shorten the viewing time of the video, playback speed of the video is increased. While time compression is quite similar to the technique of Fast Forward, it is more comprehensive.

There are various visual summarization techniques for DVLs, and by synthesizing the research that has been conducted, we decided to highlight four of them:

• Slide shows: in this technique keyframes are displayed at rapid intervals; the rate of the slide show is essential,
• Storyboards: an array of keyframes are displayed; keyframe selection is essential,
• Sequence: the more relevant clips detected from the video are assembled to maintain the flow of the story; the length of the sequence is essential.
• Time Compression: the playback speed of the video is increased. This system allows the adjustment of playback speed from 50% to 250%, in 10% increments.

For designing sufficient video abstraction UIs, both temporal and spatial presentation should be taken into consideration. Browsing video in one film frame will limit the interface to temporal presentation only. Therefore, users can look at only one point in the video at a time [11]. In order to enrich the interactivity of the system, and to mix temporal and spatial presentation, a combination of techniques can be applied. Another alternative for a better browsing solution is enabling transition between abstractions [4].

In FDVL [11] two of the techniques (storyboards and slide shows) are used together, and this is a good example of a successful combination. This kind of combination is an effective solution especially for DVLs consisting of long digital videos. In this way, the interactivity of the user interfaces and user inputs are increased.

3.2 Cultural Dimensions

People differ across linguistic and country boundaries, and user requirements are strongly influenced by the users' local cultural perspectives. There are some key theories from the field of intercultural communication, and the main goal of these theories is to decide on the cultural values for culturally diverse audiences [9], [10], [18], [21]. When these models are adapted for use in interface design, either one of the models will be preferred or more than two of the models can be synthesized. Being aware of cultural dimensions will help us to present a guideline of design heuristics for a local community.

Nowadays, cultural usability issues of DLs are an important research area [5]. Video summarization user interfaces should also be examined in detail in consideration of cultural criteria. In order to initiate a comprehensive research project, we synthesize different intercultural communication theories and select the most important cultural dimensions for designing software:

• Individualism vs. collectivism: the extent to which people are integrated in tight social networks,
• Power distance: the extent to which less powerful people expect and accept that power is distributed unequally,
• Masculinity vs. Femininity: the relative desirability of material success versus the quality of life, and of assertive versus modest behavior,
• Uncertainty avoidance: the extent to which people tolerate ambiguity and risk, or feel threatened by change,
• Context: the extent to which people in different cultures show or restrain their emotions [8].

4. Description of Proposed Research

In order to create an effective DVL design, two main levels of browsing should be taken into consideration: 1. browsing the collection (overview UIs) and 2. more detailed browsing of a digital video (preview UIs) [15], [19]. Comparing four different approaches of user information-seeking behavior, Lee and Smeaton [11] identified five information-seeking stages with regard to DVLs (see Fig. 1.). For easy and quick retrieval of the target video, effective UIs should be designed for each of the information-seeking stages. Visual summarization of the video is applied in third and fourth stages of the process.

Fig. 1. Information seeking stages for DVLs

4.1 The Comprehension of the Research

In our research we will concentrate on one part of the complete DVL cycle: the process that takes place from getting search results to deciding on the desired video. In his effort for redesigning OVDL, Geisler [7] brings up a four-stage conceptual model: gather, sift, evaluate and use (see Fig. 2). In our study, we are concentrating on the sift and evaluate stages. In the sift stage you basically get an overview of the search results, adjust the size and ordering of visible set, and select a video. In the evaluate stage you examine video abstractions, and evaluate quality and content of a video. In this stage you can also compare your choice with other similar videos.

After searching or browsing within a video archive, the UI of search results appears, showing more then one video result. You can shift between layout options and display thumbnails of the videos, and can feature metadata in different sizes and details. Within this level, by selecting one of the videos, you can jump to the UI of video details. Before passing to the video previews, you can also shift to the video-comparing user interface, where more than two videos can be compared. This gives you the opportunity to find similar videos and make more comprehensive and quicker decisions.

Fig. 2. Conceptual model of the Research

The main design issues of our research are video summarization user interfaces, and within the evaluation level, you can shift between different visual abstraction techniques. For increasing the interactivity of the DVL, we suggest a new user interface approach whereby the video(s) can be edited manually. In this way, you can re-edit or summarize videos manually and save the result. Although there are some copyright issues to be considered, we suggest an option for composing more than one video and creating a new edition. We believe these opportunities will challenge the entertaining part of the project and will increase user interaction with the system.

4.2 Research Problem

Our approach to the field of DVLs is to analyze ongoing projects and synthesize their data. We also examined the cultural usability field in detail and decided on the cultural attributes on which we should focus. In our research we will try to carry out a multidisciplinary study: we will try to integrate the cultural aspects into a DVL design process. We will concentrate on video summarization UIs and conduct usability tests for aesthetic satisfaction and design efficiency. Therefore, the research question of our study is: "Is localization a criteria for aesthetic satisfaction of user interfaces in digital video libraries?". And "if localization is a criteria, what are the specific challenges for the Turkish User?"

4.3 Details of the Turkish Advertiser Foundation Video Archive (TAFVA)

The Turkish Advertiser Foundation Video Archive (TAFVA) that will be used for creating video summarization user interfaces and for conducting usability tests includes over 1,500 advertising films that have been broadcast on Turkish TV channels and on cinema screens over the last 18 years. The length of the video films varies between 5 seconds and 5 minutes, and most of the films are in color and contain sound. Most of the videos in the TAFVA are short films (between 30 seconds and 1 minute) and they are in the genre of TV commercials.

Our research is the first step for creating a DVL of the TAFVA collection. There are two main goals of the project:

• providing a place for inspiration, information and entertainment for the people of the advertisement world,
• providing a place for education, research and entertainment for students and researchers from different universities.

5. Brief Description of the Methodology Adopted

In our study, different UIs for visual summarization of digital videos will be designed, and the Turkish user's aesthetic satisfaction behaviors will be tested. For defining the goal of the study, usability tests will be conducted for different visual summarization techniques and for different user interface designs, and test results will be analyzed and evaluated.

In order to decide on the cultural characteristics of the Turkish users, an initial research study will be carried out. Usability experts will conduct heuristic evaluation tests according to cultural dimensions (see Sect. 3.2), and cultural characteristic of the Turkish user will be explored. In addition, the cultural attributes (methaphors, colors, color combinations, navigation tools, interaction principles, etc.) for a local user interface will be determined, and finally, a guideline for design of a graphical user interface for the Turkish user will be prepared.

The TAFVA will be used for designing video summarization UIs for local users. A prototype will be designed according to already prepared guidelines, and the evaluation of the system will be conducted in a laboratory environment by comparing the designed UI to the UIs of other DVL projects (see Sect. 3.1). For usability testing the performance measurement method offered by Nielsen [16] will be performed, and two usability attributes (efficiency of use and subjective satisfaction) will be measured on a minimum of 150 novice and 50 expert Turkish users.

Before starting the design of the UIs, usability metrics of interest to the project will be discussed, and the relative importance of usability attributes will be determined. At the end of the usability testing, quantitative data of the research evaluations will be obtained, and results will be compared.

6. Current Status

As of the time of writing this paper, the TAFVA has been analyzed; analog videos of the archive have been digitalized, the metadata have been prepared and technical issues of the GUI have been discussed. The biographical base of the project has almost been completed, and the methodology of the project has nearly been decided upon. Initial research of the study is being carried out. Two usability studies that will provide valuable information about the characteristics of the Turkish User are nearly finished.

7. Proposed Solution(s)

At the end of the research project, we are willing to define a specific guideline and/or to explore some specific rules explaining how to build a layout of a video summarization interface, comparing global user feedback with that of specific user groups. We also expect to complete a summarization interface design approach for Turkish people and to define a local design base study for further academic research and industrial applications.

References

1. <http://pc-erato2.iei.pi.cnr.it/echo>, last visited 25 September 2007.

2. <http://www.informedia.cs.cmu.edu/>, last visited 25 September 2007.

3. <http://www.open-video.org>, last visited 25 September 2007.

4. Christel, M.G., Conescu, R.M.: Addressing the Challenge of Visual Information Access from Digital Image and Video Libraries. Joint Conferance on Digital Libraries. Denver, Colorado, USA (2005).

5. Duncker, E., Theng, Y.L., Mohd-Nasir, N.: Cultural Usability in Digital Libraries: Bulletin of the American Society for Information Science: April/May 26:4 (2000), pp. 21-22.

6. Geisler, G.: Agile Views: A Framework for Creating More Effective Information Seeking Interfaces. Doctoral Dissertation, University of North Carolina (2003).

7. Geisler, G: The Open Video Project: Redesigning a Digital Video Digital Library: American Society for Information Science and Technology Information Architecture Summit. Presentation. Austin, TX, February, (2004).

8. Gould, E.W.: Synthesizing the Literature on Cultural Values. In: Aykin, N. (ed.): Usability and Internationalization of Information Technology. Lawrence Erlbaum Associates, Mahwah New Jersey (2005), pp. 79-122.

9. Hall, E.T.: Beyond Culture. Anchor Press, Garden City New York (1997).

10. Hofstede, G.: Cultures and Organizations: Software of the Mind. McGraw-Hill, New York (1997).

11. Lee, H., Smeaton, A.F.: Designing the User Interface for Fischlar Digital Video Library. Journal of Digital Information 2 (2002).

12. Li, F.C., et.al.: Browsing Digital Video: CHI. The Hague, Amsterdam 2:1 April (2000), pp. 169-176.

13. Marchionini, G.: A Briefing on the Evolution and Status of the Open Video Digital Library. International Journal on Digital Libraries 4 (2004), pp. 36-38.

14. Marchionini, G., Geisler, G.: The Open Video Digital Library. D-Lib Magazine 8 (2002) <http://www.dlib.org/dlib/december02/marchionini/12marchionini.html>.

15. Marchionini, G.., Plaisant, C., Komlodi, A.: Interfaces and Tools for the Library of Congress National Digital Library Program. Informational Processing & Management 34 (1998), pp. 535-555.

16. Nielsen, J.: Usability Engineering. Academic Press, San Diego CA (1993).

17. Savino, P., Peters, C.: ECHO: a Digital Library for Historical Film Archive. International Journal on Digital Libraries 4 (2004), pp. 3-7.

18. Schwartz, S.H.: Beyond Individualism/Collectivism: New Cultural Dimensions of Values. In: Kim, U., Triandis, H.C., Kagitcibasi, C., Choi, S-C., Yoon, G.. (eds.): Individualism and Collectivism: Theory, Method, and Applications. Sage, Thousand Oaks CA (1994), pp. 85-119.

19. Shneiderman, B.: Designing the User Interface: Strategies for Effective Human-Computer Interaction. 3rd edn. Addison Wesley Longman (1998).

20. Smeaton, A.F., Lee, H., McDonald, K.: Experiences of Creating Four Video Library Collections with the Fischlar System. International Journal on Digital Libraries 4 (2004), pp. 42-44.

21. Trompenaars, F., Hampden-Turner, C.: Riding the Waves of Culture: Understanding Diversity in Global Business. McGraw-Hill, New York (1998).

22. Wildemuth, B.M., et al.: How Fast Is Too Fast? Evaluating Fast Forward Surrogates for Digital Video. Joint Conference on Digital Libraries (2003).

Appendix

Metin Çavus is a doctorate candidate in the Art and Design Doctorate Program at Yildiz Technical University, Istanbul, Turkey. He is a researcher in the Art and Design Faculty of the same university and a member of Interactive Media Design Research Group.

This leading research group of Turkey aims to investigate new dimensions in interaction design concerning current development in technology, and global and local user expectations. Our interdisciplinary approach aims to bring together the energy of people from diverse research areas.

The study of Metin Çavus is one of the first research projects in the field of DVL in Turkey, and issues like cultural usability and localization are of high importance for our future perspectives.

© Copyright 2007 Metin Çavus

Top | Contents
Guest Editorial
Next Article
Home | E-mail the Editor