Analyzing Video Data from a Field Study--论文代写范文精选

这篇essay代写范文区分了这样的对话机制,我们认为这种区别是相关的交互设计,因为他们必须意识到引人注目的社会规范。下面的essay代写范文进行讲述。

Abstract
A field study with a simple robotic companion is being undertaken in three iterations in the framework of a EU FP7 research project. The interest of this study lies in its design: the robotic interface setup is installed in the subjects' homes and video data are collected during ten days. This gives the rare opportunity to study the development of human-robot relationships over time, and the integration of companion technologies into everyday life. This paper outlines the qualitative inductive approach to data analysis, and discusses selected results. The focus here is on the interactional mechanisms of bringing conversations to an end. The paper distinguishes between "closing" as the conversational mechanism for doing this, and "closure" as the social norm that motivates it. We argue that this distinction is relevant for interaction designers insofar as they have to be aware of the compelling social norms that are invoked by a companion's conversational behaviour.

INTRODUCTION 
Much research and development is dedicated to virtual (agents) and embodied (robotic) devices that should serve as assistive technology especially to older or homebound people. One interesting field of application for agents and robots is their use as an interface to smart rooms and homes. Conversational interfaces would be the most natural and easy way to communicate with them, but they raise users' expectations in their scoial capabilities [1]. They should • be aware of the user's world, both physical and social • respect social norms and individual habits • be useful and helpful where needed without indebilitating the user • gain social status so that advice is accepted and all that not just once, but day after day. Practical experiences with today’s conversational interactive systems show the deep gap between the utopian companion and today’s clumsy attempts, see for example [2, 3, 4]. We are still far away from what, taken together, we could call "sociability", the skill, tendency or property of being sociable or social, of interacting well with others, involving elements of culture, situation, status, identity, task, communication, emotion, personality, and body, all at once and as an integrated whole. New insights are being gained in nearly each of the domains relevant to sociability individually (cf. [5]), but they will not add up to sociability automatically. Nor do we know in what way sociability with robots is specific and distinct. 

The project SERA (Social Engagement with Robots and Agents) has set out to study how systematic progress towards sociability of companions can be made. We realized that we need to know more about how users of companions actually interact with them in the real world, i.e. their homes, and over a period that is long enough to let novelty effects wear off. Apart from toys like Pleo or Aibo, most interactive robots and ECAs have not left the lab yet. The majority of tests and evaluation studies with them are done in short, task-oriented interactions. The studies of the FitTrack system [6, 7] are a notable exception, in that they comprised several weeks of continuous use of the system with its embodied agent Laura, and in that the system was intended to render real service to the user (as contrasted with toy robots or chatbots). However, the FitTrack system is PC-based, and its use is completely set apart from the user's everyday world and activities. Kidd's [21] study of the AUTOM robot is similar to ours in several respects, but is focussed more on the outcome (weight loss) than on the interaction with the robot.

THE FIELD STUDY: SCENARIO, SETUP AND DATA COLLECTION 
Like FitTrack and AUTOM, the SERA field study is being carried out in the context of potential companion applications in health and fitness assistance, especially for older people [8, 9]. In contrast to them, however, the primary goal of the study is to collect data on people's everyday life with such a robot, and not to evaluate the system. So, while the application scenario of monitoring the subjects' physical activities is based on real application interests in rehabilitation, here it mainly served as a cover story for creating enough interaction situations for the study. Furthermore, the field study does not serve the purpose of directly informing the design of a "real" companion application. The robot was set up intentionally as a purely experimental system: the goal of the project is not to improve a specific system, but to gain new insights into the requirements for and designs of companions for long-term use. The assumption underlying this approach consequently is that it is indeed possible to gain insights into humancompanion relationships that can be, to a degree yet to be established, generalized from one specific system to a broader range of future companion architectures. 

The way in which this is done in the project is to have results of the field study inform the definition of a reference architecture. This as well as the field study being work in progress at the time of writing this paper, it is not yet possible to assess the validity of the underlying assumption. Given this approach, we decided to use a low-cost experimental system which largely depends on commercially available components. The hardware consists of a (hidden) desktop computer with broad-band mobile internet connection. The periphery consists of the Nabaztag, a rabbit-like WIFI device marketed as a slightly animated talking Internet interface, (see http://www. nabaztag.com/en/index.html), a passive infrared (PIR) motion detector, a micro-switch on a hook for the house keys, a webcam (for data collection) and the array micro phone which will be used in later iterations for speech recognition (see Fig. 1). Speech recognition could not be implemented in the first iteration from which the data discussed here originate because the uncontrolled, noisy test environment and input led to intolerable latencies and recognition rates [22]. Instead, yes/no buttons had to be used as input device at this stage. This first iteration served mainly as a testbed for the experimental setup, so we did not plan for a wide range of different interactions or prolonged dialogs. Surprisingly, the study provided data that were more interesting than expected.

The participants were told that the goal of this study was to help users lead a healthy lifestyle. The background is provided by the goal to have companions assist patients recovering from a heart attack [8]. The dialogs therefore centered around the activity plan that the participants had established beforehand in the preparatory interviews with the researchers. The companion asked the participants about and reminded them of the activity plan, they were asked how they felt at the end of the day and to reflect whether they had had the right amount of activity. They were also asked whether they had weighed themselves that day as part of the fitness scenario. The Nabaztag could also provide participants with a weather report and could pass on messages from the researchers. 

The Nabaztag initiated interactions at five different occasions, namely (1) at the first appearance of participants in the morning, (2) when participants were going out of the house, (3) when participants were coming home (4) at a designated time after the last planned activity of the day and (5) when a message from the researchers had been received. Cases (2) and (3) were triggered by the participant removing/putting back the house keys on the key hook, (1), (4), and (5) were triggered by the motion sensor signal. Video data were collected on a voluntary basis: subjects were invited to start the recording every time an interaction was initiated. The users were briefed about the companion and encouraged to interact with it, but were otherwise free in their choice of when, how, and how often to do so [10]. The first iteration involved three subjects who had the setup installed in their homes for about 10 days. 65 video recordings were collected, of which 45 could be used for analysis. They are, roughly, one minute in length on average. As one can see from the still (Fig. 2), the camera angle behind and/or above the Nabaztag shows the participants not only in close-up while interacting with it, but also in their movements around the room, passing by, or going out/coming in.

DATA ANALYSIS AND METHOD 

Data were distributed among the project partners who undertake their analysis in parallel, using both quantitative and qualitative methods [11]. As laid out above, the primary goal was neither to evaluate the qualities of the system with a view to its improvement, nor the effects on users' actual health-related behaviour and lifestyle, but to enrich the body of knowledge about people relating and interacting with companion technologies. The approach to data analysis is therefore explorative in nature, to which qualitative methods are deemed particularly appropriate. Furthermore, the small sample (which is being increased for the following iterations of the field study, but never to numbers which would allow for serious quantitative analysis across subjects) is an additional factor that invites qualitative analysis.

Video interaction analysis is, to date, rather a collection of methods and practices [23] than a well-defined methodology for the empirical investigation of the interaction of human beings with each other and with (objects in) their environment. It investigates human activities such as talk, nonverbal interaction, and the use of artifacts and technologies, identifying routine practices and problems and the resources for their solution. Its roots lie in ethnography (especially participant observation), sociolinguistics, ethnomethodology, conversation analysis, kinesics, proxemics, and ethology [12].

51Due网站原创范文除特殊说明外一切图文著作权归51Due所有；未经51Due官方授权谢绝任何用途转载或刊发于媒体。如发生侵犯著作权现象，51Due保留一切法律追诉权。(论文代写)
更多essay代写范文欢迎访问我们主页 www.51due.com 当然有essay代写需求可以和我们24小时在线客服 QQ:800020041 联系交流。-X(论文代写)