Video needs at the different stages of television program making process

Transcription

1 Markkula, M. & Sormunen, E. (2006). Video needs at the different stages of television program making process. The First Symposium on Information Interaction in Context (IIiX), Copenhagen, Oct 18-20, Video needs at the different stages of television program making process Marjo Markkula and Eero Sormunen University of Tampere, Department of Information Studies University of Tampere, Finland {Marjo.Markkula, Abstract. The paper reports a field study on the needs for archive video in different stages of the television program making process. The work process typically consists of six basic stages: idea generation, planning, shooting, preselecting, script writing and editing. During the stages, the journalist gradually develops the program idea into a detailed script. The needs for archive video also evolve with the stages towards increasing specificity. The high level of uncertainty at the planning stage often leads to intense and unfocused searching of the archive and to a collection of a pile of textual listings of video material potentially useful in the task. The pile of materials at hand was consulted frequently especially in lengthy productions like documentaries. The perceived availability of archive and shot video obviously affected the content of the final script. A special limitation of the conventional archive and one reason for maintaining the pile of textual listings is that the journalist often sees the audiovisual contents of retrieved videos very late in the work process. The paper discusses the potential changes in the journalistic work process and the needs for archive video in the case of integrated video retrieval systems supporting instant access to audiovisual contents in addition to textual annotations. 1. Introduction Television broadcast archives are a challenging application for the developers of integrated multimedia search systems. The application is expected to efficiently process enormous volumes of video data, to automate labor-intensive indexing and annotation processes, to offer effective means to query for and navigate in video materials and to be well integrated into the complex editorial production system. The first integrated TV broadcast archives became operational already in the late 1990s [2,5] but many TV broadcasters are still working with conventional technology. Conventional video archives are based on a text database of annotated video documents and a mechanically operated repository of video tapes [8]. The state-of-the-art technology in video IR does not make it possible to automate intellectual archival processes but helps to improve the productivity and quality of cataloguing, indexing and annotation work. Most importantly, integrated video retrieval systems entail major improvements in searching functions [5,17]. In the conventional video archive, searching is a time consuming two-step procedure. The

2 2 Marjo Markkula and Eero Sormunen user first searches the database containing textual surrogates of the video. To view the audiovisual content she has to order the video tapes of interest from the manually operated repository and browse the tape by replaying it on a special device [8, 19]. In integrated video retrieval systems, the user has instant access to video surrogates, for example in the form of storyboards or slide shows. [4, 22]. The mainstream of research on video retrieval has focused on algorithmic developments. Algorithms or prototype systems have also been exposed to controlled user evaluations (see e.g. [3,10]). The new evaluation forum for video IR, TRECVID especially, has activated the use of controlled user evaluations in research [18]. Unfortunately, field studies on the requirements, needs and behavior of real users are still very rare. The views on user needs for video IR have remained very sparse in terms of research [17]. Task-based information seeking and retrieval is a growing branch of research [11,21]. However, the authors found only one study where researchers had collected empirical data on the use of video archives in a genuine work task situation. 1 This is regrettable since the use of video archives is very often inextricably linked with the primary work task. Replacing a conventional, two-step video retrieval system with an integrated retrieval system is a radical change in the work environment. It may change workflows in the newsroom, the searching behavior of individuals or collaboration between journalists and archivists [19]. For a general description of the television program production, see [14]. In this paper we examine how video material is used in the course of the daily work of television program making. The study deals with video needs and searching by television professionals journalists, broadcast assistants, image journalists and information specialists. Our approach is work task oriented. We focus primarily on the analysis of work tasks and the team performing them rather than on individual users of the archive. The main research aim was to learn how television programs are made and how the archived material is used in the work process. In this paper we report our answers to the following questions: 1. What are the basic stages of the work process and their characteristics? 2. What are the typical uses of archive video? 3. What kind of video needs are connected to the process and how do the needs change in the course of the work process? The paper is organized into the following structure. Section 2 contains a review of related research. Section 3 presents a description of the research setting and methods used in data collection and analysis. In Section 4 we report the results of the study: (1) description of the work process in six stages, (2) main types of video uses, and (3) the analysis of video needs in each of the work task stage. Conclusions are drawn and discussed in the final section. 1 Tan, E. (Ed.) (1998). End user requirements. Current practice, current use and envisioned use of television archive content. Technical report no. VICAR-T1.2-VUA Amsterdam: Vrije Universiteit. Unfortunately, we have not succeeded in obtaining a copy of the report. We have used the summary of the study in Tan & Müller 2003.

3 2. Related work Video needs at the different stages of television program making process 3 Tan and Müller [19] investigated if and how digitalization changed tasks and roles in television archives from the system development point of view. The study derived from an earlier study on European national television archives. All archives were conventional consisting of a reference database and a manually operated repository of video tapes. The study outlines the program production process in which archive materials are used. The process was reported to consist of seven stages: (1) script writing, (2) shooting, (3) collecting archive material: searching and first selection, (4) final selection of archive material, (5) program assembling, (6) final editing and (7) documentation. Tan and Müller found that video searching in the archives was a two-step process. First, the material was searched for in the reference database, textual annotations were studied and interesting reference records selected for later use. Second, video tapes or film reels identified in the first step were ordered and viewed before the final selection. The steps were often performed by different individuals. The process was found to be time-consuming and costly, since film reels and tapes required a lot of manual processing. Many candidate tapes were ordered because selection was impossible on the basis of text surrogates [19]. Tan and Müller predicted that digital archives will change work-tasks and workflow in television program production and archiving. Digitalization allows (1) direct searching of video material, (2) re mote and direct access to the physical copies of the video and (3) integration of production and archiving tasks. However, they point out that the main advantage of digital archives instant browsing of audiovisual contents can only be exploited if browsing and inspection functions are fast and intuitive. Requests received by video archives were analyzed by Herzum [9] and Sandom & Enser [15]. Unfortunately, requests (and search logs) are a poor source of research data if we want to understand the real needs of the user and how the needs are associated with the underlying work activity. Yang and Marchionini [23] explored what kind of relevance criteria video users apply in searching and selecting videos. They interviewed four professional video users, who were asked to recall one work-related situation in which they needed to find videos. The relevance criteria were categorized into three main categories: textual, visual and implicit criteria. Topicality was found to be the most important criterion in video selections. The participants used the textual metadata - topicality, date and authorship - as starting points and used visual criteria for final selections. The participants reported that it was hard to get any sense of the video without seeing it. This study demonstrates how to use real professional users as assessors of video materials, but its shortcoming is that it does not anchor data collection to authentic work task situations.

4 4 Marjo Markkula and Eero Sormunen 3. Research setting, data and methods The study was conducted at YLE, which is Finland's national public service broadcasting company. 2 The data was gathered by interviews. Some observations were made to achieve familiarity with the work of the program teams studied, e.g., observing the morning meetings and cutting sessions in the editing room. 3.1 Subjects The data was collected at YLE in (1) documentary, (2) factual, (3) current affairs and (4) news production departments. In total, 17 subjects were interviewed: 9 journalists, three image journalists, three broadcast assistants and two information specialists. Four of the journalists worked mainly on documentaries, one for a current affairs magazine program and three for factual magazine programs. However, one of the journalists in factual programs had also worked on news and documentaries and another one was currently engaged in documentary production. Thus, one journalist in the data might talk about various program types. Three image journalists were working on news and searched for video material for journalists needs. They helped journalists in developing illustration ideas, and also selected video and edited news inserts. They also maintained the newsroom s own video collection. Two of the broadcast assistants worked for current affairs and one for a factual magazine program. Among various other duties the assistants took care of searching for and acquiring archive video material in response to journalists requests. The two information specialists searched for information and video material for journalists in various departments. Both were intensively engaged in documentary making processes as the members of a production team. However, a substantial part of their daily work was to catalogue and annotate video materials for the program archive. 3.2 Data collection and analysis Time -line and theme interviews were primarily used as the methods of data collection. In the time-line interviewing method introduced by Dervin [6], the respondent is asked to recall and describe her own experiences of some situation as a sequence of events and then to discuss various aspects of the events in detail. Interviewees were first asked to describe a work task in which they needed or searched for archive video material. After that they were asked to describe the events in the process in detail. Some journalists also showed samples of their program(s). The journalists time -line interviews typically took a lot of time and they described only one time -line. For example, broadcast assistants might describe three time-lines in a similar period of time. 2

5 Video needs at the different stages of television program making process 5 In the theme interviews, the participants were asked to discuss themes determined separately for each professional group. Subjects were asked questions related to their work tasks and video needs, searching and selection, likewise their opinions on the textual documentation and retrieval system. Some of the participants were both timeline and theme interviewed. The numbers of subjects and interviews are presented in Table 1. The interviews were taped and transcribed. Inductive content analysis was used for data analysis (see e.g. [16]). To understand and create an overview of the program making process, the timelines were divided into phases based on the actions and intentions of the interviewees in accomplishing their task. The time -lines were then compared to each other and a generalized model of the program making processes could be outlined. Even though there were minor differences between program types (the phases were often given diffe rent emphasis in different program types), they followed the same general practice of program making. Video needs and uses related to the phases were analyzed and categorized. The theme interviews were used to supplement and validate the picture of the work processes and video needs. Table 1. The number of subjects and interviews. User group & department Number of subjects Interviews Number of time-lines Number of theme interviews Total (n) Journalists documentary factual current affairs news Broadcast assistants factual current affairs Image journalists, news Information specialists Total (n) YLE video archives and collections The main archive, YLE s broadcast archive preserves all broadcast domestic TV programs. Some work tapes have also been archived, likewise some old Finnish films. Searching is done through a reference database. The actual film reels and tapes in various formats are preserved in a vault. Cataloguing and annotation of contents is performed partly by broadcast assistants who are members of the program production 4 A comprehensive personal report on work processes, video needs and uses written for the project by a television journalist at YLE, working in news and current affairs.

6 6 Marjo Markkula and Eero Sormunen team and partly by information specialists working in the archive. The extent and style of content descriptions varies widely between production teams. Sequences assessed as important are described in greater detail than others. The content description is time coded and includes, for example, the overall topic of the program, a more or less detailed description of what happens in the image, the names of interviewees and the language and locations at which the interviews were made. The titles of source programs of the archive video clips used are always documented. Searching in the reference database is possible at the levels of broadcasts and inserts. Search results may be shown in various formats from titles to full records. Searching is available online to end-users throughout the company. Some of the journalists searched the database regularly themselves, some rather delegated video searches to broadcast assistants, image journalists or information specialists. The first evaluation of the retrieved video material is based solely on the textual descriptions. Textual records were typically printed out if assessed as interesting on screen. At some point in the process a decision was taken as to what video tapes were ordered from the manually operated repository for viewing. Some departments, teams and even individuals had their own archives. For example, the newsroom video collection maintained by the image journalists was heavily used. The image journalists selected raw video material shot by journalists or shooting teams making news reports. The archive is not permanent, older material is destroyed and the tapes reused. The existence of local video collections is an interesting phenomenon and this should be considered when integrated video archive systems are planned and implemented. Table 2. Features of different TV program types in the data. Work time frame Program duration Dominance of current news events in topics Program format Sources used for archive video Domestic news stories, cables Current affairs inserts Factual inserts Documentaries hours week three weeks month to years cables s stories ca min. dominate, some times commentaries wellestablished news room collections, regional exchange, broadcast archive 2,5 8 min. 6-8 min min. some, also human interest stories flexible to some extent broadcast archive, news room collections, international exchange not much, also retrospective stories flexible broadcast archive not important, many retrospective or historical very flexible broadcast archive, other archives incl. people s personal archives

7 Video needs at the different stages of television program making process Program types In this paper we define the television program from the production viewpoint as a unitary piece of document (work) created by a journalist for a broadcast. This definition is important, since many broadcast programs as seen by the TV viewer are collections of separately prepared and topically detached units (e.g., cables in a news broadcast). The different departments of YLE work on different types of programs. In our data, we found four program types which have quite disparate characteristics. Some of the differences between the program types described by the subjects are summarized in Table 2. In this paper we do not focus on the effects of program types on archive video needs. Table 2 is presented to show that our data is rich in variation regarding program making processes. 4. Findings 4.1 Work processes of television program making Figure 1 illustrates the common work process of television program making derived from the interviews. The work process is divided into six stages: (1) idea generation, (2) planning, (3) shooting, (4) pre-selecting video, (5) script writing and (6) editing. The program idea and the script are processed and developed throughout the work process. Video needs may arise and videos may be searched for at all stages of the process. At first the volume of video material acquired (or paper prints of video descriptions) increases. The material is pruned throughout the process but much pruning takes place in the later stages of the work process when the actual videos are assessed. The program making process began with idea generation. The output of this task was the decision to make or not to make the program. According to the interviewees, ideas for programs may come from various sources. Sometimes idea generation occurred at team meetings. Sometimes the journalist developed the program idea quite alone. Background work was usually needed before the decision about the program was made and the work began. Especially in documentaries, extensive searching for video was common at this stage, but it happened in smaller scale in other program types, too. The actual work began at the planning stage. The outline for the program script was written at this stage. The planning stage was characterized by information and material gathering. This might include reading material, interviewing experts or talking to colleagues. People and places were sought for shooting. Searching in the video archive was also very active at this point.

8 8 Marjo Markkula and Eero Sormunen Work stock of video (archive/shot) Idea Script Outline Final Stages of the work process Idea Planning Shooting generation Preselection Script writing Edit Time Fig. 1. Stages of the work process in making a television program. The shooting was planned in the preceding stage by the journalist and carried by the journalist and a shooting crew. For most programs, some or most of the video material needed was shot. After shooting, additional archive video might be searched for. The next step was pre-selection, where video material shot for the program and/or acquired from the archives was often browsed at one go. The relevant video clips were pre -selected, given time -codes and commented. Some journalists compiled selected clips on disc, some on tape. Some documentary journalists used an editing table for pre-selection since it allowed playing around with the video clips. Supplementary video searches in the archive might be carried out. The script writing of the program was completed based on the material acquired. Usually separate scripts were made for text and for video image. The text script included the text for the news reader, speaker or voice-over. The video script was sometimes very detailed including, for example, transition proposals (e.g. fade-outs), sometimes more general leaving much work to be done at the editing stage. The edit was done according to the script by the editor and the journalist in the editing room. The editors have a major role in program making and most journalists seemed to highly respect their expertise. They are skilled not only in handling the technical device but also have eye for detail such as composition. The journalist explained her ideas to the editor, who tried to implement them. In the edit, the final selection of video material for the program was made. The edit started by digitizing the material selected in the pre-selection stage. The journalists tended to take much more video material (three or four times as much according to one journalist) to the editing room than actually needed in order to have some options to choose from. In documentaries, new video searches were still made. For other program types, the schedules seldom allowed this. The boundaries of the stages are fluid and different work tasks may sometimes be carried out concurrently. For example in news, the tight schedules might force the journalist to do one task on top of another. In current affairs and factual programs, the production process seemed to be quite tightly scheduled necessitating rigorous organization of tasks. The journalists considered the process of documentary making to be more flexible than other program types.

9 Video needs at the different stages of television program making process Video uses The subjects could name various uses for archive videos which can be categorized into two main groups. Videos were used (1) as information sources to obtain information to support idea generation, planning and decision making and (2) as material in comp iling the program. This distinction is important and is based on the Buckland s typology of the meaning of the word information including information-as-process (i.e. a change in knowledge), information-as-knowledge (i.e. what is imparted in information-as-process), and information-as-thing (i.e. objects capable to inform) [1]. Fidel [7] has also made the same distinction by introducing the notions of data pole and object pole in using images. The use of an image is close to the data pole if it is primarily an information source, for example, in solving a problem. The use of an image to communicate something, for example, to characterize a phenomenon in a report, is located close to the object pole. Videos were needed as information sources, i.e. information-as-knowledge, (1) to generate and develop program ideas, (2) to find information and check facts, (3) to find people for interviews in the planned program. Archive video is perceived primarily as raw material in the program making process. We could identify three different ways of using video-as-thing: 1. The most common practice was to use only the image component of a video to illustrate something. Illustrations aimed to concretize, emphasize, enliven or exemplify the topics read by the news caster or voice-over or referred to by the interviewee. For example, according to an image journalist, in cables, typically 20 seconds of illustrative video (archived or newly shot) and seconds of newsreader is shown. Illustrative video usually supports spoken text. For example, an interview with a medical expert talking about the dangers of using various medications concurrently was illustrated by a hand opening a drug locker, close-ups on drug packages etc. 2. The use of a complete video clip was also common. All media components of the video were included in the program. For example, people s statements (e.g. politicians) were used to recall or substantiate what had been said. Clips from old movies were used to add humour or lighten the subject. Video was seen as a quick way to impact the historical continuity of some topic. 3. The audio track of an archive video was sometimes used alone, for example, to demonstrate the voice or speaking style of a person to enhance her portrait. 4.3 Needs for video at different work stages Video needs may arise and video material may be acquired many times during the program making process (see Fig. 1). The most extensive video acquisition typically takes place at the idea generation, planning and shooting stages.

10 10 Marjo Markkula and Eero Sormunen At the idea generation stage, the videos were primarily used as information sources, even though some of the material could later be found useful as program material. Often the purpose of video searching was to find out what had been done earlier on the topic. For example, a journalist recalled her request to an information specialist in the following way: I am planning a program on XL -models. What has been done on the topic, how and when? Complete programs and inserts were looked for. The journalists aimed to find out if the program idea was worth developing and if a new viewpoint could be found on the topic. Further, the availability of video material for the envisaged program was checked. At this stage, the journalist often worked solely with the textual descriptions of videos. Documentary journalists wished to see video material at this point, at least if they were not familiar with the subject area. At the planning stage, archive video was often gathered in large volumes to be sure of having at least some useful material. The journalist did not yet know in detail what the program would be like, what exact topics would be discussed, and what kind of material could and would be shot. As one journalist put it: well the first list is off the cuff. Information and facts were also searched for. Documentary journalists looked for potential interviewees. Videos were usually searched for or requested several times during the planning stage: the journalist s knowledge of the topic increased, the script outline and the video needs became more focused. A broadcast assistant described this in the following way: Often it goes like they call all day, that now I need and it just occurred to me and could you check if there is anything on At the shooting stage, the journalist s primary concern was the new material to be shot. Interviews and also material to illustrate the interviews were shot, if possible, at the same shooting location. After shooting, the journalist knew what material had been shot and what might still be needed from the archives. For example, a journalist needed a video of a turnip field. The insert was made in winter, when it was impossible to shoot such material. At the pre-selection stage, new video needs arose, when the journalist actually saw all the acquired video material in a single session. Until this point, she had mainly worked with the text ual descriptions. Sometimes the available video was found to be insufficient. A documentary journalist gave an example of this kind of a situation: I need something to illustrate this incident when the policemen came to his door the departure of a police car out to the street or maybe just some street scenes of the city in the 1950 s. In documentary making as well as in other longer work processes, additional video material was still looked for. However, it was common that the journalist had to be prepared to manage with the video material at hand. This was one reason why much larger volumes of video material than actually needed were searched for at the planning stage. At the script writing stage, no examples of archive video needs could be found in our data. According to the journalists, the script is usually adapted to the limits of the material available just assessed at the pre-selection stage. Urgent needs for additional material have already been identified there. This assumption is supported by an image journalist working with video material for news, who argued that sometimes a journalist may come with a readymade script and ask for a specified illustration

11 Video needs at the different stages of television program making process 11 already written in the script. This was considered troublesome, since envisaged video matching the script is often difficult to locate quickly enough. At the editing stage, video needs often concerned illustration and the images needed could be described in great detail. Editor is often the first person together with the journalist to see the whole program and may thus identify problems, for example, in the flow of the narrative. In editing, video needs may arise when o two clips do not look good together and a third clip is needed between the two o a segue from one topic to another or from one interviewee to another does not work o an interview has to be shortened and the cuts hidden with illustrations o a bus, train etc. is going into the wrong direction in the video at hand o something is missing in the narrative. An example of such a specific need was given by an information specialist: We need a few seconds of sunset above the sea. The sun is filtering through the leaves in the beach. A feeling of a paradise isle. 5. Conclusions and discussion In this paper, we introduced a process model of television program making. The work process was observed to consist typically of six, partly overlapping stages: (1) idea generation, (2) planning, (3) shooting, (4) pre-selecting the video material, (5) script writing and (6) editing. We also analyzed what types of video needs are typical for different stages of the work process and how the needs change with the stages. Our model of the stages of the TV program making process is analogous to the model for information searching process (ISP) by Kuhlthau [12]. Kuhlthau developed and tested her model in longitudinal studies on high school and college students working on extensive essays (term papers). The goal of writing essays is learning and the end-product is a mere proof of things learned. Program making is a professional activity where the primary goal is the end-product. Our subjects perceived a similar uncertainty regarding the goals and outcomes of the task at the early stages of the process as emphasized in the ISP model. Our view of the work process differs somewhat from the process outlined by Tan & Müller in [19]. We found that script writing was not a separate subtask at the beginning of the process. The program idea was developed through all stages and the detailed script was written after the collected material had been assessed and preselected. Further, our analysis revealed that video needs arose and videos might be acquired throughout the process even though the most active searching takes place at the beginning. Typically, video material was gathered in the first half and pruned in the latter half of the work process. However, new material was searched also in the latter half but the volumes of video retrieved were limited. Two main categories of archive video use were identified. Initially, videos were primarily used as information sources. In the later stages video clips or their parts were used mainly for compiling the program. We also found that video needs became more focused towards the end of the work process. The obvious reason for the

12 12 Marjo Markkula and Eero Sormunen increased focus is the gradual elaboration of the program idea into a detailed script. However, it should be noted that even in specific needs (read: a specific illustration problem) the user may accept very different pieces of video. As reported in a study on journalistic illustration tasks by Markku la & Sormunen [13], different types of images can be fluently assigned to a given context. The findings are also in line with longitudinal studies on ISP, where needs for information change from general and vague towards specific and directed after the focus of the task at hand has been fixed, see [12,20]. The interplay between the evolving program idea and the cumulative stock of material acquired seems to characterize the journalistic work process in television. It is likely that the needs to browse and collect video material in creating ideas and planning the program will persist even when more interactive video retrieval technologies are used. In the integrated video archive system, the journalist has much better tools to test program ideas since the audiovisual contents can be browsed while searching. This might speed up testing of program ideas. Further, the planning stage could become more effective since the team could much more easily estimate the availability of material useful for the task at hand. But even more, the advantages of instant access to audiovisual contents might facilitate solving specific illustration problems at the later stages of the work process. It remains an open question how much the improved interaction with video surrogates will reduce the need to collect a cumulative stock of potentially useful material in the early stages of the process. A corresponding study in a digital photo archive in a newspaper found that the journalists tended to collect possible images early in the writing process and make the final selection quite late when the page layout was designed [13]. This suggests that integrated video retrieval systems should offer a collaborative tool for the team to retain the stock of interesting materials during the program making process. Here we have presented overall findings related to the television program making process and video needs. The data collected is also valid for a more detailed analysis of the two-phase search process. An interesting future analysis will be the relevance criteria emerging from the work task situation. The compulsory two-phase search and selection process might help in identifying the differences between selections based on textual descriptions and on audiovisual content. The further analysis will also shed light on the requirements for textual annotations and video surrogates. Differences observed in video needs and uses between program types will also be analyzed in detailed. References 1. Buckland, M. (1991). Information and information systems. New York; Westport, Connecticut; London: Praeger. 2. Buscher, I. (1999). Digital video archives facing the facts. EBU Technical Review, No. 280, Available: [downloaded 17 April 2002]. 3. Christel, M.G. & Conescu, R.M. (2005). Addressing the Challenge of Visual Information Access from Digital Image and Video Libraries. In: Proceedings of the 5th ACM/IEEE-CS

13 Video needs at the different stages of television program making process 13 Joint Conference on Digital libraries. New York, ACM Press, p Available: [downloaded 17 April 2002]. 4. Christel, M. & Warmack, A. (2001). The effect of text in storyboards for video navigation. In: Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP '01). Available: [downloaded 17 April 2002]. 5. Del Pero, R., Dimino, G. & Stroppiana M. (1999). Multimedia catalogue the RAI experience. EBU Technical Review, No. 280, Available: [downloaded 17 April 2002]. 6. Dervin, B. (1992). From the mind s eye of the user. Sense-making qualitative-quantitative methodology. In J. D. Glazier & R. R. Powell (Eds.), Qualitative Research in Information Management (pp ). Englewood (CO): Libraries Unlimited. 7. Fidel, R (1997) The image retrieval task: Implications for the design and evaluation of image databases. The New Review of Hypermedia and Multimedia, 3, Green, E.L. & Klasén, L. (1993). Indexing and information retrieval of moving images experiences from a large television information database. In: Raitt, D.I. & Jeapes, B. (Eds.) Online information 93 : Proceedings of the 17th International Online Information Meeting. Oxford: Learned Information, p Herzum, M. (2003) Requests for information from a film archive: a case study of multimedia retrieval. Journal of Documentation 59 (2) Hollink, L, Nguyen, GP et al. (2005). Assessing User Behaviour in News Video Retrieval. IEE proceedings on Vision, Image and Signal Processing 152(6), Available: [downloaded 17 April 2002]. 11.Ingwersen, P. & Järvelin, K. (2005). The turn. Integration of information seeking retrieval in context. Dordrecht: Springer. 12.Kuhlthau, C. C. (2004). Seeking meaning. A process approach to library and information services. 2nd Ed. Westport, CT: Libraries Unlimited. 13.Markkula, M. & Sormunen, E. (2000). End-User Searching Challenges Indexing Practices in the Digital Newspaper Photo Archive. Information Retrieval 1(4): Newby, J. (1997). Inside broadcasting. London: Routledge. 15.Sandom, C. and Enser P.G.B. (2003) Archival moving imagery in the Digital Environment. In: Anderson, J., Dunning, A. & Fraser, M (Eds.) Digital resources for the humanities London: Office for Humanities Communication, Kings College, Schamber, L. (2000) Time-line interviews and inductive content analysis: their effectiveness for exploring cognitive behaviors. JASIS 51(8): (2000). 17.Smeaton, A.F. (2004). Indexing, browsing, and searching digital video. In: Cronin, B. (Ed.), ARIST 38(): Smeaton, A.F., Over, P & Kraaij, W. (2004). TRECVID: evaluating the effectiveness of information retrieval tasks on digital video. In: Proceedings of the 12th annual ACM International Conference on Multimedia, p Available: [downloaded 17 April 2002]. 19.Tan E. and Müller H. (2003) Integration of Specialist Tasks in the Digital Image Archive. In Cognition in a Digital World, Ed. Here van Oostendorp, Lawrence Erlbaum Associates, Inc., Publishers, New Jersey 20.Vakkari, P. (2001). A Theory of the task-based information retrieval process: a summary and generalisation of a longitudinal study. Journal of Documentation 57 (1), Vakkari, P. (2003). Task-based information searching. In B. Cronin (Ed.) Annual Review of Information Science and Technology 37. Medford: Information Today, Wildemuth, B. M., Marchionini, G., Wilkens, T., Yang, M., Geisler, G., Fowler, B., Hughes, A., & Mu, X. (2002). Alternative surrogates for video objects in a digital library: users' perspectives on their relative usability. In: Proceedings of the 6th European Conference on

14 14 Marjo Markkula and Eero Sormunen Research and Advanced Technology for Digital Libraries. Lecture Notes in Computer Science 2458, Springer, p Yang, M. & Marchionini, G. (2004). Exploring Users' Video Relevance Criteria -- A Pilot Study. Proceedings of the Annual Meeting of the American Society of Information Science and Technology, pp Nov , Providence, RI. Acknowledgements. We would like to thank all people at YLE who supported our project. Especially, we thank Juho-Pekka Rantala for his report and discussions at the beginning of the project. His help was invaluable in orientating to the practices of television program making. The study was mainly funded by the Academy of Finland in projects no and We thank the members of the FIRE group for constructive comments.