On the Design of Intelligent Memory Functions for Virtual Meeting Places: Examining Potential Benefits and Requirements Version 11


Download 100.31 Kb.
Size100.31 Kb.
1   2   3   4   5   6


The work reported here is been carried out in the context of the project Magic Lounge. The project name also stands for a virtual meeting place in which the members of a geographically dispersed community can come together to carry out joint and goal-directed activities. For example, to plan a travel they may jointly explore information sources on the World-Wide Web (WWW), such as time tables of flights and public transportation services, an on-line travel agency, hotel information, cultural events, local sightseeing suggestions, travel planning tools etc. Several Lounge meetings may be necessary to accomplish the travel-planning task that we consider eminently representative of complex activities in terms of subtasks, communication situations.

Unlike many other approaches (e.g., for chat rooms on the web or CSCW platforms) we do not assume that the users enter the Lounge via exactly the same communication channels. Rather, we imagine a scenario as illustrated in Figure 1. There, one of the user has a fully-fledged standard multimedia PC, another has a handheld PDA with a small and only medium resolution screen for the display of text and graphics, while the third user connects with a mobile phone that - apart from the audio channel - has a tiny LCD display to allow the display of short text messages and minimalist graphics.

A virtual meeting place such as the Magic Lounge can serve as a collective memory of an electronically connected community. The collective memory can be thought of as an information system based on a continuously emerging pool of information and knowledge that is brought in by the community members. To be able to use such an information system effectively, it must provide functions for the collection, structuring, conservation, distribution, and reuse of informative contents. Not only should a virtual meeting memory enable to retrieve information which was brought in by the participants and by external resources, but it should also make implicit links in exchanged messages become more explicit for the user querying the memory.

The concrete implementation of such functions, however, is strongly guided by the particular purposes for which it will be used. For example, especially large enterprises show an increasing interest in gaining the capacity for so-called organisational memories. They are motivated by the insight that productivity can be increased, if the employees could only draw on the past experiences of their colleagues, rather than making again the same or wasting effort in the development of already seen solutions. At least successful companies usually have some understanding of their business processes as well as the flow of work and information. Furthermore, companies often have already certain guidelines for the format and media in which they represent the relevant information to build their own organisational memory. For the general case, there is no such concrete domain. This makes it much more difficult to propose memory functions in more detail. Though we have identified the joint planning of travel as a possible sample scenario within the Magic Lounge project, we do not aim at the development of a kind of specialised virtual traveller's corner or even travel agency.

Figure 1: Several users virtually meet in the Magic Lounge while having different configurations. The two users at the left-hand side access via PC’s with audio and high-resolution graphics. In contrast, the users on the right hand side have to cope with media limitations of their access devices. The PDA device has only low-resolution graphics and no audio channel. The mobile phone supports audio but its tiny display only allows to display minimalist graphics and telegram-style text messages.

Figure 1 : Plusieurs utilisateurs ayant à leur disposition diverses configurations matérielles se rencontrent dans Magic Lounge. Les deux utilisateurs de gauche ont accès à des PCs offrant des moyens audio et graphique à haute résolution. Les utilisateurs de droite, en revanche, possèdent des interfaces limitées : l’Assistant Personnel n’offre que du graphique à basse résolution et ne comporte pas de canal audio ; le téléphone mobile possède une connexion audio mais son écran de petite dimension ne permet d’afficher que du graphique simplifié et des messages textuels en style télégraphique.

As our work is directed towards the development of a general platform for virtual meeting places, we focus on the provision of memory functions which are more or less independent of a particular subject matter such as the business world or travel planning. We concentrate on the information that emerges from observing communications among the users. That is, communicative acts, such as speaking, typing text messages, exchanging images or posting around references to documents, together with the context and the way how such contributions are made become the main source of information upon which a memory is built. However, it is important to point out that our notion of a conversation memory for a virtual meeting place is not just that of a large “container” or database in which to collect and store information that may become critical to conversational performance at a later point in time. Rather we argue that informative memory contents requires an active creation process involving both the memory as source of information and the user who wants to satisfy particular information needs. Following this view our research interest focuses on methods and interfaces for constructing meaningful memory contents at retrieval time.

In the following we sketch a few fictive scenarios to illustrate how the inhabitants of virtual meeting places might benefit from memory services. For the purpose of illustration the scenarios draw on three fictive persons, Susan, Lars, and Hank, who once met by chance in the Magic Lounge, eventually teamed up and since then usually enter the Lounge after leaving off work to see what's going on there. Throughout the scenarios we will use the abbreviation VMM to refer to a Virtual Meeting Memory.



Susan, Lars and Hank know each other from a number of previous meetings in the Magic Lounge. As they have become friends over the time, they plan to meet in the real world too, and therefore try to find a good place to do so. Susan and Lars already started to work on this joint travel-planning task a few days ago, unfortunately without Hank who was on a business trip. Today Hank is also connected to the Magic Lounge. When entering the Lounge, he gets informed about who else is there, and the other inhabitants of the Lounge are made aware of Hank being there. Susan greets Hank and informs him that she is searching the web for a meeting place. Furthermore, she tells Hank that he may consult the VMM to get informed what she and Lars have discovered and discussed during the last sessions. Though not designed especially to support travelling planning, the VMM provides Hank with information that allows him to get an overview on the current state of the planning task at hand. For example, Hank may query the VMM which messages were exchanged between the inhabitants of the virtual meeting place, and use such email traditional functions as sort these items according to various criteria such as the temporal order of occurrence, the sender or addressee(s), or the type of media (e.g., text, voice, graphics, video); or such as plain text search on the content of exchanged messages. Also, the VMM keeps track of information units, which are brought in from external sources such as a web page. Similar to message records, Hank can browse through the record of these units, too. However, there are more advanced memory functions, which will help Hank to get an overview much faster. By virtue of intelligent document classification, message extraction, and summarisation functions, the VMM allows Hank to overview the types of the messages without displaying them again. Moreover, due to filtering mechanisms the VMM is able to "clean up" its contents in certain ways. For example, a user navigating through the World-Wide Web often finds her/himself circling around certain pages. Even when Hank is interested in the details of a past meeting, there is usually no need to have him repeat such circles in order to give him an idea which web pages were visited during that meeting. To sum up, it is important to note that Hank can choose between several different views when accessing memory contents. Some of these views are basic in the sense that they do not require deep processing of the information units, e.g. a view that just lists contributions to a conversation on a time axis. Other views on memory contents, however, can only be realised if contents and intention of a message are understood to a certain extent by the system.


While the first scenario intended to illustrate how a VMM can be exploited by a "late-comer", the second one focuses on intelligent assistance for Magic Lounge participants during a meeting, i.e., while they communicate with each other or perform activities such as web surfing. For example, suppose Susan has a web browser, which can be controlled using speech input. A spoken command such as "Show me again the web page with the hotels in London that Hank has found" can only be successfully executed by her browser when the memory of the meeting (or even a past meeting) is taken into account. Besides the mere provision of information on demand, however, the VMM can take on a more active role, too. There are often situations in which a user would appreciate a helpful comment or a hint without asking for it explicitly. For example, while Susan clicks herself from one web page to the other, the VMM notices that Hank has visited some of these pages just a few minutes before. In order to make Susan aware of this fact, and to provide her with the additional option to follow Hank's navigation path through these web-pages, her browser now changes the appearance of exactly those hyperlinks which were selected by Hank and displays a brief message to inform her about the situation and the new option. In this case the VMM provides information for the recognition of a situation in which a suggestion would be appropriate as well as the trace of Hank's previous navigation path. After a while, Susan finds a list of some recommended restaurants, and would like to know which of those Hank and Lars would prefer. Unfortunately, Lars has already left the Lounge. While discussing to which restaurant they will go, the Magic Lounge again recognises that there is a situation in which the conversation partner should be provided with a comment. This time, it reminds Susan and Hank about Lars preference for Asian food that he stated in an earlier meeting. Further services of this kind include remembering stated arguments, previously made agreements as well as remembering the participants' goals that have been inferred from their communications and actions.


For the purpose of the third scenario, let’s assume Lars is walking through a shopping promenade while being connected with the Lounge through his mobile phone. While walking along the street, he passes a travel agency that offers a cheap flight and accommodation package to a place he would like to go with his friends. Lars takes the chance to inform his friends about the information he found. In response to this, Susan and Hank both sitting at their PC’s at home start a search on the World-Wide Web in order to check how good the offer actually is. Later at home, Lars will not only find all the references to the pages found by his friends but will also be able to trace back the conversation they had on the issue. It is a major goal of the Magic Lounge project to enable and facilitate the collaborative exchange of information in a collaboration process that is achieved via heterogeneous communication media. Thereby, the concept of having a memory is essential to support such heterogeneous collaborations through time and situations.

Download 100.31 Kb.

Share with your friends:
1   2   3   4   5   6

The database is protected by copyright ©ininet.org 2024
send message

    Main page