Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Look up keyword
Like this
2Activity
0 of .
Results for:
No results containing your search query
P. 1
Visualization of Mined Pattern and Its Human Aspects

Visualization of Mined Pattern and Its Human Aspects

Ratings: (0)|Views: 193 |Likes:
Published by ijcsis
Researchers got success in mining the Web usage data effectively and efficiently. But representation of the mined patterns is often not in a form suitable for direct human consumption. Hence mechanisms and tools that can represent mined patterns in easily understandable format are utilized. Different techniques are used for pattern analysis, one of them is visualization. Visualization can provide valuable assistance for data analysis and decision making tasks. In the data visualization process, technical representations of web pages are replaced by user attractive text interpretations. Experiments with the real world problems showed that the visualization can significantly increase the quality and usefulness of web log mining results. However, how decision makers perceive and interact with a visual representation can strongly influence their understanding of the data as well as the usefulness of the visual presentation. Human factors therefore contribute significantly to the visualization process and should play an important role in the design and evaluation of visualization tools.
Researchers got success in mining the Web usage data effectively and efficiently. But representation of the mined patterns is often not in a form suitable for direct human consumption. Hence mechanisms and tools that can represent mined patterns in easily understandable format are utilized. Different techniques are used for pattern analysis, one of them is visualization. Visualization can provide valuable assistance for data analysis and decision making tasks. In the data visualization process, technical representations of web pages are replaced by user attractive text interpretations. Experiments with the real world problems showed that the visualization can significantly increase the quality and usefulness of web log mining results. However, how decision makers perceive and interact with a visual representation can strongly influence their understanding of the data as well as the usefulness of the visual presentation. Human factors therefore contribute significantly to the visualization process and should play an important role in the design and evaluation of visualization tools.

More info:

Published by: ijcsis on Sep 09, 2009
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

03/12/2011

pdf

text

original

 
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 4, No. 1 & 2, 2009
Visualization of Mined Pattern and Its HumanAspects
 
Ratnesh Kumar Jain, Dr. R. S. Kasana
Department of Computer Science and ApplicationsDr. H. S. Gour Central UniversitySagar, MP, India. jratnesh@rediffmail.com, irkasana7158.gmail.com
Dr. Suresh Jain
Department of Computer Engineering, Institute of Engineering & Technology,Devi Ahilya University, Indore, MP (India)
 
suresh.jain@rediffmail.com
 Abstract
—Researchers got success in mining the Webusage data effectively and efficiently. But representation of the mined patterns is often not in a form suitable for directhuman consumption. Hence mechanisms and tools that canrepresent mined patterns in easily understandable formatare utilized. Different techniques are used for patternanalysis, one of them is visualization. Visualization canprovide valuable assistance for data analysis and decisionmaking tasks. In the data visualization process, technicalrepresentations of web pages are replaced by userattractive text interpretations. Experiments with the realworld problems showed that the visualization cansignificantly increase the quality and usefulness of web logmining results. However, how decision makers perceiveand interact with a visual representation can stronglyinfluence their understanding of the data as well as theusefulness of the visual presentation. Human factorstherefore contribute significantly to the visualizationprocess and should play an important role in the designand evaluation of visualization tools. This electronicdocument is a “live” template. The various components of your paper [title, text, heads, etc.] are already defined onthe style sheet, as illustrated by the portions given in thisdocument.
 
 Keywords-Web log mining, Knowledge representation,Visualization, Human Aspects..
I.
 
I
NTRODUCTION
The dictionary meaning of visualize is "to form a mentalvision, image, or picture of (something not visible or present tosight, or of an abstraction); to make visible to the mind orimagination" [The Oxford English Dictionary, 1989]. Thediscovery of Web usage patterns would not be very usefulunless there are mechanisms and tools to help an analyst betterunderstand them. Visualization has been used very successfullyin helping people understand various kinds of phenomena bothreal and abstract. Hence it is a natural choice for understandingthe behavior of Web users. “The essence of InformationVisualization is referred to the creation of an internal model orimage in the mind of a user. Hence, information visualization isan activity that humankind is engaged in all the time”. [1]
Figure 1. Visualization Process
Visualization of the web usage data is a technique in whichmined data are represented graphically. In this process,technical representations of web pages are replaced by userattractive text interpretations.
 A.
 
VISUALIZATION TECHNIQUES
There are a large number of visualization techniques whichcan be used for visualizing the data. In addition to standard2D/3D-techniques, such as x-y (x-y-z) plots, bar charts, linegraphs, etc., there are a number of more sophisticatedvisualization techniques (see fig. 2). The classes correspond tobasic visualization principles which may be combined in orderto implement a specific visualization system.
Figure 2. Classification of Visualization technique
Comp.Rep. of realityReality User(s) Picture(s)
48ISSN 1947 5500
 
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 4, No. 1 & 2, 2009
1)
 
Geometrically Transformed Displays
Geometrically transformed display techniques aim at finding“interesting” transformations of multidimensional data sets.The class of geometric display techniques includes techniquesfrom exploratory statistics, such as scatter plot matrices andtechniques which can be subsumed under the term “projectionpursuit”. Other geometric projection techniques includeProjection Views, Hyperslice, and the well-known ParallelCoordinates visualization technique..
Figure 3. Parallel Coordinate VisualizationFigure 4. Dense pixel displays (Courtesy IEEE)
2)
 
 Iconic Displays
Another class of visual data exploration techniques is theiconic display techniques. The idea is to map the attributevalues of a multidimensional data item to the features of anicon.
3)
 
 Dense Pixel Displays
The basic idea of dense pixel techniques is to map eachdimension value to a colored pixel and group the pixelsbelonging to each dimension into adjacent areas. See figure 4.
4)
 
Stacked Displays
Stacked display techniques are tailored to present datapartitioned in a hierarchical fashion. In the case of multidimensional data, the data dimensions to be used forpartitioning the data and building the hierarchy have to beselected appropriately. See figure 5.
 B.
 
 INTERACTION AND DISTORTION TECHNIQUES
In addition to the visualization technique, for an effective dataexploration, it is necessary to use some interaction anddistortion techniques. Interaction techniques allow the dataanalyst to directly interact with the visualizations anddynamically change the visualizations according to theexploration objectives and they also make it possible to relateand combine multiple independent visualizations. Distortiontechniques help in the data exploration process by providingmeans for focusing on details while preserving an overview of the data. The basic idea of distortion techniques is to showportions of the data with a high level of detail, while others areshown with a lower level of detail.
Figure 5. Dimensional Staking display (Courtesy IEEE)
1)
 
 Dynamic Projections
The basic idea of dynamic projections is to dynamicallychange the projections in order to explore a multidimensionaldata set. A classic example is the GrandTour system [24],which tries to show all interesting two-dimensional projectionsof a multidimensional data set as a series of scatter plots.
2)
 
 Interactive Filtering
In exploring large data sets, it is important to interactivelypartition the data set into segments and focus on interestingsubsets. This can be done by a direct selection of the desiredsubset (browsing) or by a specification of properties of thedesired subset (querying). Browsing is very difficult for verylarge data sets and querying often does not produce the desiredresults. Therefore, a number of interaction techniques havebeen developed to improve interactive filtering in dataexploration. Examples are Magic Lenses [26], InfoCrystal [27]etc.
3)
 
 Interactive Zooming
In dealing with large amounts of data, it is important to presentthe data in a highly compressed form to provide an overviewof the data, but, at the same time, allow a variable display of the data on different resolutions. Zooming not only means todisplay the data objects larger, but also means that the datarepresentation automatically changes to present more detailson higher zoom levels. The objects may, for example, berepresented as single pixels on a low zoom level, as icons onan intermediate zoom level, and as labeled objects on a highresolution. Examples are: TableLens approach [28], PAD++[29] etc.
4)
 
 Interactive Distortion
Interactive distortion techniques support the data explorationprocess by preserving an overview of the data during drill-down operations. The basic idea is to show portions of the datawith a high level of detail while others are shown with a lowerlevel of detail. Popular distortion techniques are hyperbolicand spherical distortions, which are often used on hierarchiesor graphs, but may be also applied to any other visualizationtechnique. An example of spherical distortions is provided inthe Scalable Framework paper (see Fig. 5 in [23]). Other
49ISSN 1947 5500
 
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 4, No. 1 & 2, 2009
examples are Bifocal Displays [30], Graphical Fisheye Views[31] etc.
5)
 
 Interactive Linking and Brushing
The idea of linking and brushing is to combine differentvisualization methods to overcome the shortcomings of singletechniques. It can be applied to visualizations generated by allvisualization techniques described above. As a result, thebrushed points are highlighted in all visualizations, making itpossible to detect dependencies and correlations. Interactivechanges made in visualization are automatically reflected inthe other visualization. Typical examples of visualizationtechniques which are combined by linking and brushing aremultiple scatterplots, bar charts, parallel coordinates, pixeldisplays, and maps. Most interactive data exploration systemsallow some form of linking and brushing. Examples arePolaris [22], XGobi [25] and DataDesk [32].Experiments with the real world problems showed that thevisualization can significantly increase the quality andusefulness of web log mining results. However, how decisionmakers perceive and interact with a visual representation canstrongly influence their understanding of the data as well asthe usefulness of the visual presentation. In section III we tryto explore the human aspects in visualization. In section IV wediscuss some research examples.II.
 
R
ELATED WORK
 Most common technique of visualization is Graph drawingand it has been subject of research since decades [5, 9].Graphs are a natural means to model the structure of the web,as the pages are represented by nodes and the links representedby edges. Many graph algorithms are used, in original oradapted form, to calculate and express properties of web sitesand individual pages [4, 7, 8]. Although to a lesser extent,graph theoretic methods have also been applied to the usernavigation paths through web sites [10]. WebQuilt is a loggingand visualization system [11] which is interactive in the sensethat it provides semantic zooming and filtering, given astoryboard. Webviz [2], VISVIP [3], VisualInsights [12] aresome other visualization tools. So many commercialvisualization tools for representing association rules have alsobeen developed. Some of them are MineSet [14] and QUEST[13]. Becker [15, 16] describes a series of elegant visualizationtechniques designed to support data mining of businessdatabases. Westphal et al. [17] give an excellent introductionof visualization techniques provided by current data miningtools. Cockburn and McKenzie [6] mention various issuesrelated to graphical representations of web browsers’revisitation tools.How a viewer perceives an item in a visualization displaydepends on many factors, including lighting conditions, visualacuity, surrounding items, color scales, culture, and previousexperience [18]. There are so many technical challenges indeveloping a good visualization tool one of the big challengesis
User acceptability.
Much novel visualization techniqueshave been presented, yet their widespread deployment has nottaken place, because of user acceptability due to lack of visualanalytics approach. Many researchers have started their work in this direction. An example is the IBM Remail project [20]which tries to enhance human capabilities to cope with emailoverload. Concepts such as “Thread Arcs”, “CorrespondentsMap”, and “Message Map” support the user in efficientlyanalyzing his personal email communication. MIT’s projectOxygen [19] even goes one step further, by addressing thechallenges of new systems to be pervasive, embedded,nomadic, adaptable, powerful, intentional and eternal. Usersare an integral part of the visualization process, especiallywhen the visualization tool is interactive. Rheingans suggeststhat interaction should not be simply a “means to the end of finding a good representation” [21]. Interaction itself can bevaluable since exploration may reveal insight that a set of fixed images cannot. Human factors-based design involvesdesigning artifacts to be usable and useful for the people whoare intended to benefit from them. Unfortunately, thisprinciple is sometimes neglected in visualization systems.III.
 
H
UMAN
F
ACTORS
 How people perceive and interact with a visualization tool canstrongly influence their understanding of the data as well asthe system’s usefulness. Human factors (e.g. interaction,cognition, perception, collaboration, presentation, anddissemination) play a key role in the communication betweenhuman and computer therefore contribute significantly to thevisualization process and should play an important role in thedesign and evaluation of visualization tools. Several researchinitiatives have begun to explore human factors invisualization.
 A.
 
Testing of Human Factors
There are so many Human Computer Interaction interfacesavailable. Each interface is tested for its functionality(usability study) and ease of interaction (user studies).
1)
 
 Ease of interaction
To test ease of interaction we consider only real users andobtain both qualitative and quantitative data. Quantitative datatypically measures task performance e.g. time to complete aspecific task or accuracy e.g. number of mistakes. User ratingson questions such as task difficulty or preference also providequantitative data. Qualitative data may be obtained throughquestionnaires, interviews, or observation of subjects using thesystem.Walenstein [45] describes several challenges with formaluser studies. According to him the main problem in the userstudies is that we studies so many users but the true facts aboutthe ease and benefits can be told only by the experts who canbe difficult to find or may not have time to participate inlengthy studies. Another problem is that missing orinappropriate features in the test tool or problems in theinterface can easily dominate the results and hide benefits of the ideas we really want to test. Thus, it seems that userstudies can only be useful with an extremely polished tool sothat huge amounts of time must be invested to test simpleideas that may not turn out to be useful. One solution to thisproblem is to have user studies focus on design ideas rather
50ISSN 1947 5500

You're Reading a Free Preview

Download
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->