A User-interaction Approach in an Information-cockpit

Christian Kaiser

, Selver Softic

, Vedran Sabol

and Mario Zechner

Virtual Vehicle Research and Test Center, Graz University of Technology, Inffeldgasse 21A, Graz, Austria

Know-Center GmbH, Graz University of Technology, Inffeldgasse 13, Graz, Austria

Keywords:

Visualization, Faceted Search, Metadata, Semantics, Future Work Place.

Abstract:

This position paper introduces an approach for user interaction in an information-cockpit. The information-

cockpit in our case is an explorative visual user interface backed by a faceted search-engine. The user interface

employs several visualizations helping the user to get an overview of the data and providing intuitive drill-down

functionality on search results. Besides searching and visualizing of data on a personal drive, our information-

cockpit supports federated searching where ﬁles are located across heterogeneous data sources in the network.

1 INTRODUCTION

In the last years a massive growth of online and of-

ﬂine data often lead to an information overload of the

user. Structuring of the information stored in differ-

ent data sources greatly supports searching, but even

when guidelines do exist, interpretations of the im-

posed structures are likely to vary. Therefore, users

who are not familiar with the system face a serious

challenge when they need to search for and gather

information. In a future workplace scenario a user

should not have to deal with such problems - the sys-

tem should take over the task of presenting the infor-

mation in a uniﬁed, understandable way. This idea is,

to a certain degree, comparable to an online search:

instead of scanning or searching different Web-pages

to ﬁnd information, search engines are used as a sin-

gle point of access. A real-life example would be a

project-manager that is supposed to collect informa-

tion from several sources for a milestone report. In

this case the manager needs an overview about activ-

ities of his engineers instead of detailed knowledge

about every line of code as well as a single access

point to all the information. Our information cockpit

represents a uniﬁed visual point of access providing

search and visualization of information stored in het-

erogeneous data sources. In this work the focus is

on the design of the visual user interface, since this

has been identiﬁed as a key success factor for faceted

search user interfaces (Hearst, 2006). Our concept is

based on visualizations offered by available visualiza-

tion libraries, such as D3.js Data-Driven Documents

JavaScript library (Bostock, 2013).

2 RELATED WORK

According to Hearst (Hearst, 2009; Hearst, 2008;

Hearst, 2006) the success of a user interface for search

is very highly dependent on its design. One important

aspect is that people prefer to be allowed to navigate

and ﬁlter along multiple metadata ﬁelds, which pro-

vides them with a sense of control over the data. In

case of visual support for search the work of Kien-

reich (W. Kienreich and Seifert, 2008) and Smith

(G. Smith and D.S., 2006) as well as the book of

Fill (Fill, 2009) have to be mentioned. Smith states

that the point where a simple folder hierarchy for

documents is sufﬁcient to organize, keep track and

search is already long past. However, many of the

existing systems providing search and browse capa-

bilities are still textual. Given ongoing explosion

of gathered content enabled by automatic data col-

lection mechanisms, a text-based interface may be-

come increasingly overwhelming to users, expecially

novices (G. Smith and D.S., 2006). Kienreich intro-

duces visual representations for ﬁltering search re-

sults along topical, geographic and temporal dimen-

sions. Tvarozek (Tvarozek and Bielikova, 2008)

shows the usage of novel navigation and visualiza-

tion approaches for search visualization such as visual

depiction of facets and restrictions, visual navigation

and graph like exploration of search results. Polowin-

ski (Polowinski, 2009) states that faceted search is

particularly suitable for structured but heterogeneous

data with explicit semantics.

635

Kaiser C., Softic S., Sabol V. and Zechner M..

A User-interaction Approach in an Information-cockpit.

DOI: 10.5220/0004510906350639

In Proceedings of the 9th International Conference on Web Information Systems and Technologies (FWP-2013), pages 635-639

ISBN: 978-989-8565-54-9

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

3 DATA PREPARATION

The data used for this example has its origin in the

”Tankia” project of the TU Graz racing-team, an in-

ternational engineering competition of student groups

that compete in the formula student. In this section

the preparation of the data is brieﬂy described. The

data is a mixture of ofﬁce documents and ﬁles created

by special engineering tools. Before the data can be

indexed by the search engine it is necessary to ﬁrst ex-

tract searchable information from the ﬁles and store

it in a pre-deﬁned format. A data preparation pro-

gram was implemented for extracting useful informa-

tion and metadata, and storing the results in XML

format. After that it is possible to build a Web-service

with a REST (Representational State Transfer)-API

(Application Programming Interface) which provides

data for given questions. A JSON

(JavaScript Object

Notation) formatted response of the API is preferred.

The following code illustrates a possible JSON re-

sponse.

{

"result_information": {

"#results": "500",

"facets": {

"person": "100",

"car": "300",

"security": "130"

"navigation": {

"page_id": "1",

"next_page_id": "2",

"query": "http://gui.org/search&query=

Hans&facets=person+car+security"

"results": [

{

"uri": "file://path/xy.extension",

"source": "N:",

"metadata": {

"modified": "2013:01:25:19:36:43",

"authors": "Hans Huber",

"created": "2012:12:21:15:36:43",

"type": "text",

"title": "car instructions"

"snippet": "[..]<b>car</b>[..]",

"facets": [

"person",

"car"

"relevance": "0.000134342"

}

]

}

see http://www.w3.org/XML/

see http://www.w3.org/TR/json-ld-api/

For the purpose of designing the REST-API which

will be used by the visual user interface a set of 500

ﬁctive search results is deﬁned. The data structure re-

turned as a ﬁctive search result in JSON format can

be seen above (for shortness just 1 result is shown).

The extracted keywords are called facets, e.g. ”per-

son”. According to Hearst (Hearst, 2006) facets are

deﬁned as following: Facets refer to categories which

are used to characterize information items in a collec-

tion. Each facet has a name, such as person or car

for a car-sharing collection. Facets can be ﬂat or hi-

erarchical. Some navigation information is provided

to allow pagination. A query with page-id ”2” added

to the original query would lead to a response with

the next page of the result-list. The results provide

information about the unique URI, the source, some

metadata like ﬁle-created date, ﬁle-modiﬁed date, the

author, etc. Additionally a HTML (Hypertext Markup

Language) styled snippet of the occurrence of a key-

word is provided as well as the matched keywords and

a calculated relevance.

4 GRAPHICAL USER

INTERFACE (GUI)

At this point it is assumed that the data is prepared and

searchable through the REST API. The next and most

important step is the design (and later implementa-

tion) of the graphical user interface. There are two

design guidelines that we wish to follow:

• ensure that usability of the GUI is high as this is a

crucial point in regard to user acceptance.

• use different visualizations, each being adapted to

the type of faceted metadata

GUI design shown in Figure 1 should, for the most

part, be self-describing and its usage should be clear

and intuitive. In the rest of this section a motivation

and discussion of design decisions is provided.

4.1 Search

As the main goal of our GUI is searching there is a

text ﬁeld on the very top of it. The user can type one

or more keywords into the text ﬁeld an hit enter or

push the ”search”-button to submit the search query

to the system. A visually supported facetted search,

which is discussed in the next subsection, can be used

to ﬁlter the result set according to user’s criteria.

4.2 Visualized Faceted Search

As a search query can retrieve lots of results a possi-

WEBIST2013-9thInternationalConferenceonWebInformationSystemsandTechnologies

636

Figure 1: A mock-up of the GUI.

bility to narrow down and optimize the results is pro-

vided by using a faceted search (shown on left in Fig-

ure 1) - which is a well-known, proven exploration

and discovery technique (Hearst, 2008). The intent

is to use the faceted search additionally to an initial

keyword-search, but it should also be possible to start

with all existing results as base-amount for the ﬁlter-

ing. When the user clicks on a facet a new search

query is generated and sent to the Web-service, which

ﬁlters the result-set with regard to the chosen facets.

Filtering using any number of different facets is pos-

sible. As mentioned before, data type-dependent vi-

sual representations are used to provide a visually en-

hanced access to faceted metadata. The facet cate-

gories, e.g. ”ﬁle-type” and ”department”, rely on the

data indexed previously during data preparation. It is

the job of the system administrator to identify such

categories and make sure the chosen categories are

extractable. The bubbles represent the identiﬁed pos-

sibilities of each category. The bigger a bubble is,

the more results share this facet. Note that there is a

minimum bubble-size to keep the describing text in-

serted into the bubble readable. The bubbles share the

same color if they are members of the same category.

Alternative representations are used for non-textual

data, for example a star-representation for ”rating” or

a timeline (or a progress bar) for temporal informa-

tion.

4.3 Result Visualization

Whenever the search query changes, either because

of typing in a new keyword or selecting facets, a new

query is sent to the API. The response will be dis-

played in the result box (center of the display). Per

default a list is used to show the results, as this is

the most commonly used and widely known format of

presenting them. Above the result list there is a palette

with available views for visualizing the results. Per

default the list view is shown, with a blue-rectangle

around the list button showing that this view is in use

at the moment. Other result visualizations, such as the

bar chart showing aggregations, a line chart display-

ing numerical values, or a graph visualization show-

ing relations between search results will provide vi-

sual analysis capability to the user.

4.4 Collecting and Saving Results

A user might not be interested in only one single

search result, but in a current status of a group of ﬁles,

e.g. collect all necessary data for a milestone report.

Therefore a so called ”collection-box” is integrated

on the right side of the GUI. The box content is per-

sistent and does not change with new search-queries

allowing the user to collect ﬁles over time. Adding

hits from the search-result list (or visualization) to the

AUser-interactionApproachinanInformation-cockpit

637

collection works via drag and drop. The user can click

on the ”save as” button and provide a destination di-

rectory where the collected ﬁles will be saved.

4.5 History View

While a user has the possibility to copy the current

status of ﬁles as discussed in the last section, he

might need an overview of all previously used search-

queries too, e.g. to search again for the very same

by clicking on the saved query instead of retyping

the keywords and reselecting facets again. Therefore

the view of the GUI can be changed by clicking on

the ”History” tab below the text-ﬁeld. In this view

the used search queries are represented as trees. The

root element is the keyword used in the query or ”no

keyword” if no keyword was used. The child ele-

ments are all the facets where an element was cho-

sen, e.g. if only the ﬁle-extensions ”.m” and ”.pdf”

have been chosen, then the child element of the key-

word would be ”ﬁle-extension” and again the child

elements would be ”.pdf” and ”.m”. The History tab

contains a list of search-trees. If the user clicks on a

search-tree, the query will be opened in the search-

tab. The ”Overview” box on the bottom left edge

shows the current search-tree, which is saved in the

history tab.

4.6 Authentication

If ﬁles are used to save knowledge to a hard-drive, op-

erating systems provide access-right handling. Pre-

deﬁned users and groups either have permission to

read/write/execute or not. The approach described in

this paper works with unstructured data, e.g. ﬁles of

a network drive, and has to respect permissions. A

simple way to do that is to force users to login with

an account to authenticate. The question is, if a user

should be able to see at least some metadata of the

information which might solve his problem to get an

hint whom to ask, even if he has not the permission

to view the ﬁle. The underlying problem is that a user

might be able to gather all critical informations out of

the metadata, e.g. the snippet, and therefore can read

a ﬁle he should not. Access is a complex problem

and therefore not solved in the GUI approach yet, but

a hint is given that there should be some management

by placing the drop-down box on top-right which says

current user is ”Christian”.

5 SUMMARY AND FUTURE

WORK

In this paper a conceptional approach for user in-

teraction in an information cockpit is presented.

Hereby speciﬁc demands with regard to a future

workplace scenario have been considered. As an

outcome, knowledge about data-management to op-

timize data-extraction is retrievable in a retrospective.

A drill-down oriented approach is used to access data

combined along with state-of-the-art approaches like

faceted- and explorative-search. It is expected to pro-

vide easy and personalized access to relevant data for

users. Further interesting features for the described

approach were identiﬁed and will be brieﬂy described

in this section. An important feature focuses on han-

dling access rights and roles in the GUI. If there are no

access restrictions, e.g. in a personal environment or

in companies maintaining a free access policy for the

employees, no problem exists: every user can see the

same results. But if our information-cockpit is used

in a company environment with different departments

and lots of access restrictions the user shall see only

what he is allowed to see and this should also include

browsing through faceted metadata. Another part of

the future work involves additional visualizations for

browsing and analyzing the search results. Per default

search results are shown as a list, but showing them

within a hierarchy visualization, such as the ”circle

packing”

, might present an interesting option. Such

a visualization would be also applicable for navigat-

ing hierarchically organized metadata facets (on the

left side of the UI). For the search queries stored un-

der the history tab an additional interesting scenario is

proposed - a search agent: If a user searches for some-

thing but ﬁnds nothing or the results are not satisfying,

the user might subscribe to the query and be informed

when the result set changes. Finally, a heuristic eval-

uation of our design using a mock-up of the interface

will be performed. This will allow us to unveil ma-

jor usability issues and discover design inconsisten-

cies before the implementation of the GUI begins.

REFERENCES

Bostock, M. (2013). Data-driven documents.

Fill, H. G. (2009). Visualisation for Semantic Information

Systems. Gabler Verlag, Wiesbaden.

G. Smith, M. Czerwinski, B. M. D. R. G. R. and D.S., T.

(2006). Facetmap: A scalable search and browse visu-

See http://mbostock.github.com/d3/talk/20111116/

pack-hierarchy.html

WEBIST2013-9thInternationalConferenceonWebInformationSystemsandTechnologies

638

alization. In IEEE Transactions on Visualization and

Computer Graphics, Vol. 12 (5), pages 797–804.

Hearst, M. (2006). Design recommendations for hierarchi-

cal faceted search interfaces. In ACM SIGIR Workshop

on Faceted Search.

Hearst, M. (2008). Uis for faceted navigation: Recent ad-

vances and remaining open problems. In Workshop

on Human-Computer Interaction and Information Re-

trieval, HCIR, pages 13–17.

Hearst, M. (2009). Search User Interfaces. Cambridge Uni-

versity Press, New York, 1st edition.

Polowinski, J. (2009). Widgets for faceted browsing. In

Smith, M. and Salvendy, G., editors, Human Interface

and the Management of Information. Designing Infor-

mation Environments, volume 5617 of Lecture Notes

in Computer Science, pages 601–610. Springer Berlin

Heidelberg.

Tvarozek, M. and Bielikova, M. (2008). Personalized

view-based search and visualization as a means for

deep/semantic web data access. In Proceedings of

the 17th international conference on World Wide Web

(WWW08), pages 1023–1024.

W. Kienreich, E. L. and Seifert, C. (2008). Apa labs: An

experimental web-based platform for the retrieval and

analysis of news articles. In First International Con-

ference on Applications of Digital Information and

Web Technologies, pages 58–62.

AUser-interactionApproachinanInformation-cockpit

639