Information Visualisation

No complex and sophisticated visualisation technique can be used without a userfriendly user interface. The interface must not only be easy to use, but also “natural” for the most user groups and not only for specialists. The aim of this paper is to show some important goals at designing of user interfaces for information visualisation and on the other hand to present some existing ideas and also real examples for “well designed” interaction methods. INTRODUCTION Nowadays there are plenty of methods for visualisation of very different type of information. Visualization simply became a standard to manage and work with a great amount of information not because it just looks better than some numbers or texts in a list but because it is no more possible to work without it. Users have no chance to overlook big datasets and even less to find any relationships between the data. There are some criteria that can help to judge the usability of information visualisation techniques: a) How good is the integration of not fully qualified data – dataset with missing data? b) Is the manipulation of the visualized data easy? How fast can the function be reached? c) Is it possible to add new or missing data? How fast and error prone is the input? d) What kinds of queries are supported? How long does it take to get the result? e) Is the whole data set visible at once? Is the visualisation misleading? f) Can the user find patterns? a) How good is the integration of not fully qualified data? Sometimes there are some missing data for example the birthday of a person is missing in the database. Nevertheless it should be possible to search in the database. On the other hand this kind of data should also be visible in the visualisation or after a search. b) Is the manipulation of the visualized data easy? How fast can the function be called? Dynamic visualization techniques allow a manipulation and reorganization of the displayed data. That includes selecting a subset and comparing the data with each other. This action can sometimes only be done by calling some “very well“ hidden functions in a sub-sub-menu. This hinders the user from fast reorganization of data and makes the analysis very cumbersome. c) Is it possible to add new or missing data? How fast and error prone is the input? In some cases it can be important to be able to insert new data to see how the relations and the visualization will be changed by the new data. But generally the input of new data is a challenging problem because the user should be hindered from adding wrong and maybe redundant data. d) What kinds of queries are supported? How long does it take to get the result? Here we can distinguish between static and dynamic queries. In a static query you get the result after entering all parameters that the searched objects should have. Dynamic queries show the result after each entered parameter and make the search much more flexible because zero-hit searches as well as too large result sets can be prevented. Especially in environments like the Internet it is very crucial to get a fast feedback how large the result will be. For this kind of feedback it is enough to get the count of matched objects so you can make the search more special when the result set is too large or more general in the case of a zero-hit. e) Is the whole data set visible at once? Is the visualisation misleading? There are some cases where a visualisation can easily lead to misinterpretations. Examples are: The whole dataset is not visible because of the great amount of data Some data is occluded f) Can the user find patterns? Is the visualisation technique that flexible that it can be reorganized to see some patters and relations among the data? For this it is sometimes important to divide the data into smaller groups and to compare the items in the groups to each other or to compare the groups to each other. EXAMPLES The following examples show some ways to overcome current problems in information visualization. Photofinder Photofinder [Shn00-1][Shn00-2] is a photo managing software for digitalized photos with direct annotation. Photos are for many professional users worthless unless they are annotated with date, time, location, photographer, title, recognizable people etc. This is an example for a tool that makes user input easier, quicker and less error-prone. On the other hand it also has a better searching ability than the other products on the market. It is much more than an alternative for keeping your photos in envelops. Figure 1 Direct annotated photo. Direct annotation means that the user first enters the names of persons that are shown on the picture. Afterwards these names just have to be dragged over the head of the person (figure 1). Just the same you can do with other categories and information for example where or when the picture was taken. All the information is stored in a database. Figure 2 Search interface with dynamic query. The bars are updated after each change. The other important feature of Photofinder is the Boolean query and search interface that allows users to do Boolean queries visually. Because of the clearness only AND – across different attributes and OR – within an attributeare supported. The queries are not statically displayed but appear just in time they are entered. This kind of dynamic queries give much more opportunity to the user because it helps to prevent zero-hit as well as too general queries (figure 2). Missing data is displayed, as an additional value so there is no loss of information. Figure 3 Thumbnail in the multiple-display. The thumbnails of the results can be shown in a multiple display that reduces the waste of white space so much more pictures can be seen at once (figure 3). Figure 4 Scatter plot display. X-axis is assigned with the number of people, the y-axis with the rate