=Paper=
{{Paper
|id=Vol-1947/paper10
|storemode=property
|title=FaRBIE: A Faceted Reactive Browsing Interface for Multi RDF Knowledge Graph Exploration
|pdfUrl=https://ceur-ws.org/Vol-1947/paper10.pdf
|volume=Vol-1947
|authors=Luis Fuenmayor,Diego Collarana,Steffen Lohmman,Sören Auer
|dblpUrl=https://dblp.org/rec/conf/semweb/FuenmayorCLA17
}}
==FaRBIE: A Faceted Reactive Browsing Interface for Multi RDF Knowledge Graph Exploration==
https://ceur-ws.org/Vol-1947/paper10.pdf
FaRBIE: A Faceted Reactive Browsing Interface for
Multi RDF Knowledge Graph Exploration
Luis Fuenmayor1 , Diego Collarana2,3 , Steffen Lohmman3 , Sören Auer4
1
RWTH Aachen University, Aachen, Germany
2
University of Bonn, Bonn, Germany
3
Fraunhofer IAIS, St. Augustin, Germany
4
Technische Informationsbibliothek, Hannover, Germany
luis.fuenmayor@rwth-aachen.de
collaran@cs.uni-bonn.de
steffen.lohmann@iais.fraunhofer.de
soeren.auer@tib.eu
Abstract Linked Data brings numerous RDF Knowledge Graphs to the Web,
and as a result, the Linking Open Data (LOD) cloud comprises several independ-
ent – but linked – graph datasets. Faceted Browsing is a popular type of User
Interface (UI) to explore the knowledge in these RDF graphs. Due to the distrib-
uted and linked nature of RDF graphs, exploring more than one graph at a time
is a common interaction scenario in this context. However, most state-of-the-art
user interfaces allow users to browse only one RDF graph at a time. Additionally,
exploring multiple RDF graphs adds a degree of uncertainty to the user interface,
such as connectivity problems and longer query response times as well as vari-
ations in the size and semantic complexity of the retrieved data. In this paper, we
present a faceted browsing approach named FaRBIE, which enables users to ex-
plore multiple RDF knowledge graphs simultaneously. The reactive UI approach
implemented in FaRBIE aims to ‘react’ to the uncertainty imposed by querying
multiple RDF graphs in order to enhance the user experience. FaRBIE is com-
posed of reactive UI components, refined from common faceted browsing UIs
and improved by means of reactive design patterns. A component-based UI ap-
proach supports the convenient generation and reuse of the reactive elements.
1 Introduction
In the current age of Internet, data is being created at an explosive rate. However, bey-
ond simple data growth, it becomes more and more important to understand how such
data correlates. Consequently, knowledge graphs gain importance, as does the field of
Linked Data, which brings a increasing number of RDF Graphs to the Web. As a result,
the Linked Open Data (LOD) cloud is comprised of several independent – but linked
– graph datasets. User Interfaces (UIs) oriented towards the exploration of RDF graphs
become equally significant, as a means of bridging the gap between the technology and
non-technical users. Faceted Browsing is a popular User Interface (UI) paradigm to
explore and navigate the knowledge contained in RDF graphs.
Due to the distributed nature of Linked Data, exploring more than one RDF graph at
a time is a common interaction scenario. However, this common scenario lacks tailored
111
FaRBIE: A Faceted Reactive Browsing Interface for Multi RDF Knowledge Graph
Exploration
(a) Exploring DBpedia using SemFacet (b) Exploring Linked Leaks using OntoText’s UI
Figure 1: Motivating Example. A user typically requires two different UIs to explore
the RDF graphs of DBpedia and Linked Leaks.
and efficient UI support at the moment. Let us assume the following multi RDF and
distributed browsing scenario in the context of a crime investigation: during an ongoing
investigation for corruption, browsing and analyzing information coming from various
sources is one of the key steps performed by investigators. In the case of a "politically
exposed person", such as a politician, an investigator wants to explore whether or not
this politician is in any form involved or related to the Panama Papers scandal, and
at the same time retrieve additional general information about the politician. While the
Linked Leaks Dataset5 contains an RDF representation of the Panama Papers, DBpedia6
contains general information about politicians. Figure 1 illustrates how a user typically
requires two different UIs for exploring information about Mauricio Macri in two RDF
graphs. In this case, SemFacet (Figure 1a) is used to browse the RDF graph of DBpedia,
while the OntoText browser (Figure 1b) is used to explore the RDF graph of Linked
Leaks.
Additionally, the exploration of multiple RDF graphs brings new challenges at the
UI level. The UI has to deal with a higher degree of uncertainty, such as connectivity
problems and longer query response times. These are issues which are only aggravated
by the variations in the size of the retrieved data, and the complexity in the semantics
of the data, all factors which potentially have a negative effect on the usability of the
interface, ultimately progressing into a decrease of the overall user experience. State-of-
the-art approaches [2,4,6,9] are mainly designed to explore one RDF graph at a time.
Consequently, they do not address these challenges sufficiently.
In this paper, we focus on tackling the problem of exploring multiple RDF graphs
and its challenges, hence conceiving FaRBIE, a Faceted Reactive Browsing Interface.
FaRBIE implements a reactive user interface style with solid User Experience (UX)
design principles. In the context of this paper, a Reactive User Interface should be
understood as an interface that is not only capable of providing an interactive and real-
time feedback to the users, but also of adapting the interface in order to cope with the
challenges that are manifested in both retrieval times and results complexity from the
set of independent and multiple RDF graphs being queried. The main contributions of
5
http://data.ontotext.com/linkedleaks
6
http://wiki.dbpedia.org
112
FaRBIE: A Faceted Reactive Browsing Interface for Multi RDF Knowledge Graph
Exploration
this paper are as follows: i) Definition of the main challenges at the UI level when
exploring multiple RDF graphs at the same time. ii) A reactive component-based UI
approach that handles the uncertainty imposed by the intrinsic nature of RDF graphs.
iii) A proof of concept and an initial open source implementation using modern Web
UI development technologies.
2 Related Work
Faceted browsing on RDF data has been the focus of several research works. How-
ever, we believe the problem of easy-to-use interfaces for the exploration of multiple
RDF graphs is still underexplored. Additionally, most of the available tools are targeted
towards advanced users instead of non-technical lay users. Arenas et al. [2,4] intro-
duced SemFacet as a faceted search approach enhanced with Semantic Web technolo-
gies. SemFacet is capable to automatically generate the facet names and values from
metadata provided in RDF and OWL. Additionally, it uses the implicit knowledge of
the RDF graph to improve the interface by showing new facets or deriving new annota-
tions. SemFacet is based on a theoretical framework that accounts for both RDF data
and OWL 2 axioms [1]. However, the scope of SemFacet is limited to faceted search on
a single dataset.
Stadler et al. [9] presented Facete, a spatial data exploration application that can be
applied to SPARQL endpoints. The facets are automatically generated and a map view
of the spatial data is displayed to the user. The focus of Facete is on the exploration of
spatial data instead of general entity data, as in our approach. Additionally, Facete also
considers only one RDF dataset for exploration.
Stolz and Hepp [10] conducted an evaluation of the appropriateness of an adapt-
ive faceted search UI as a search paradigm for e-commerce on the Web of Data. In
their work, they present preliminary evidence of the applicability of an adaptive UI for
faceted search. However, in contrast to our reactive approach that changes the user in-
terface elements according to the semantics of the new data received as search results,
the UI adapts according to a user profile, temporal context, and diversification.
Ferré [5] presented Sparklis with the goal to combine the expressivity of formal
languages and the usability of faceted search. Sparklis enables non-technical users to
explore a SPARQL endpoint and answer complex questions based on facets sugges-
ted by the application. The facets are complemented by SPARQL queries to retrieve
answers to complex questions.
Finally, Khalili et al. [6] presented LD-R, a framework to develop reusable Semantic
Web driven UIs. Although the focus of the work is to provide a development framework,
inside its configuration it incorporates Faceted Browsing. Thus, the approach goes in the
same direction that is envisioned for FaRBIE.
3 UI Problems when Browsing Multiple RDF graphs
In the following, we will briefly describe what we identified as some of the main UI
problems when browsing multiple RDF graphs.
113
FaRBIE: A Faceted Reactive Browsing Interface for Multi RDF Knowledge Graph
Exploration
Reactiveness towards the semantics of the data (P1) The quality and amount of the
semantics of data in a multiple knowledge graph exploration scenario vary in a sig-
nificant way. We may find general concepts (e.g., Organization) to more specialized
concepts (e.g, Terrorist Organization). The UI needs to deal and react to the variety of
the semantics in the data, in the sense of preserving the semantics of the information be-
ing retrieved. Most of the current implementations [3,9] are designed with single RDF
knowledge graph exploration scenarios in mind, where flows of information in the form
of entities and attributes that originate from them are static in structure. In such cases,
designing interfaces that implement reactivity in the semantic context have received
more attention of researchers recently. However, this is unfortunately not the case when
it comes to the exploration of multiple knowledge graphs. This is due to the fact that
the flow of information in such cases is no longer static in structure – it is inherently
different, constantly growing and evolving, thus posing a challenge when it comes to
designing standardized but scalable interfaces that cope with such change. Hence, we
identified a lack of reactive approaches that tackle semantic contexts when it comes
to multiple RDF graph exploration. We attempt to tackle this problem by developing
reactive UI components, which allow for semantic, multiple RDF knowledge graph ex-
ploration while preserving the semantic contextualization of the results being obtained.
Error visibility and feedback (P2) Another potential pitfall that can be observed when
exploring multiple knowledge graphs is that of error visibility and feedback. Errors dur-
ing the browsing process need to be made clearly visible so as to provide better feed-
back to the users, in accordance to the visibility of the system status and the help users
recognize, diagnose, and recover from errors heuristics referred to by Nielsen [8]. At
the same time, the attention span time frames mentioned by Nielsen [7] should also be
preserved. These usability heuristics help safeguard that interfaces are sufficiently com-
fortable and functional when the user performs the intended tasks of a system, ensuring
further adoption of the technology. However, these principles are difficult to preserve in
the context of multiple knowledge graph exploration, due to the nature of how federated
searches operate – different queries for different knowledge graph sources, performed
in parallel, typically imply different response times for the queries as well. This con-
tributes to longer waiting times, where the user could be left without any feedback for
time spans that surpass a few seconds, causing confusion and triggering a sense of lack
of control in the users, which ultimately breaks the feedback flow between the interface
and the user. In the long run, this has the potential of becoming a fatal flaw that un-
dermines the development of the technology. We attempt to tackle this problem by the
implementation of a reactive UI component, which allows for improved visibility and
feedback of errors that may happen during the federated search process, where the user
can have full feedback as to the current system status, as well as obtaining information
about the nature of the errors that have occurred.
Minimalist design (P3) A UI for exploring RDF graphs needs to manage well the
topic of minimalism and avoid violating the aesthetics and minimalist design rules that
also form part of Nielsen’s heuristics [8]. To address these rules, we took three main
requirements into consideration when designing our approach: maximum screen space
114
FaRBIE: A Faceted Reactive Browsing Interface for Multi RDF Knowledge Graph
Exploration
usability, maximum visibility (to also contribute in solving P2), and minimal cluttering.
However, providing minimal cluttering in a screen space where there is both abundant
information to display and considerable functionality to support becomes a challenge,
especially in systems which have multi-layer navigation bars, as they either consume an
important block of screen space or become too confusing or difficult to follow for the
user. Thus, these design requirements were also carefully considered when designing
our UI components. Moreover, the implementation of reactive features in our design
further aims to contribute to minimizing the required screen space, as well as improving
the ergonomics of the interface by providing familiarity in the interface by means of
semantically enabling or disabling views or functions according to the results being
obtained during the search.
4 The FaRBIE Approach
To tackle the challenges of browsing multiple RDF graphs, and to keep up with provid-
ing non-technical users with a more enjoyable and usable experience, we propose a
reactive user interface design style. In contrast to imperative approaches (e.g., libraries
such as jQuery), a Reactive UI updates itself by reacting to changes in the data and
rendering the right components whenever such data changes occur. Reactive UIs are
component-oriented, where each component may evolve independently, facilitating re-
usability in the interface. A reactive UI component may contain not only the view but
also pieces of logic to react appropriately to the semantics of the data. Hence, we argue
that this style of UI fits well for RDF-based applications. ReactJS7 has become one of
the most popular libraries to implement this style of user interface.
4.1 Architecture
By making the user interface components reactive, we can provide developers with
the possibility of making such components interact with the users in real time, while
the queries for the multiple datasets are still running. In general, a common faceted
browsing UI can be divided into three main UI sections: i) the search box or input
section and source selection, where the user enters a keyword to start the search process
and select data sources and entity types of interest. ii) the facets section contains all the
UI elements to filter and narrow down the results; the facets are automatically generated
based on the results collected from the RDF graphs. iii) the results section shows the
entities found in the RDF graphs that match the keyword. Additionally, it provides
users with feedback concerning the current state of a multiple dataset query (whether
the search has succeeded, has failed, or is in progress). Figure 2 illustrates the FarBIE
architecture.
4.2 Logic Keeper
The Logic Keeper is not a user interface component but a component responsible for
handling the communication with the RDF graph servers (e.g., using SPARQL HTTP
7
https://facebook.github.io/react/
115
FaRBIE: A Faceted Reactive Browsing Interface for Multi RDF Knowledge Graph
Exploration
RDF Graph(s) Server
Level 1
Results Logic Keeper Search Faceted Results
Facets, Results, and Metadata generator Box Bar Container
Reactive UI Level 2
Search Box Source Entity Facet View Result Settings Sources
Selector Selector Navigation Bar Item Bar Infobox
Results Faceted Bar
Container Person Product
Facet Facet ... Level 3
Item Search Item Item
Result Item Facet Search
Range Date Level 4
View Bar Facet Item ...
...
Facet Facet
(a) FaRBIE architecture (b) UI components
Figure 2: UI Architecture. (a) FaRBIE architecture containing the Results Logic
Keeper and Reactive UI Components; (b) Reactive UI Components organized in levels
from generic to specific. The UI components can be extended and specialized in provid-
ing a better UX according to the semantics of data.
requests). It uses the keyword search input and prepares the queries to the RDF graph
servers. In this paper, we assume that the RDF graph server supports Keyword Search8 .
The Logic Keeper manages all the logic applied to the results coming from the RDF
graph servers, including 1) facets generation, 2) entity results preparation, and 3) meta-
data creation from the search process.
To generate the facets, we based our implementation on the work of Arenas et
al. [3]. In brief, generic SPARQL queries are applied on the resulting entities to generate
the facets. Additionally, the list of facets and the number of facet values are computed
and maintained by the Logic Keeper. The second responsibility of the Logic Keeper is
that of preparing the entity results, where a snippet is composed per result, and values
such as dates are standardized for the reactive UI components. Finally, meta-data about
the search process is created, including provenance of results, server request success
or error information, the number of results, the number of results by type and the type
of entities. All this metadata is computed and managed by the Logic Keeper. To trigger
the reaction of the UI components, the Logic Keeper communicates whenever there is
a change in the search results data, under the following circumstances: 1) more results
arrive from the RDF graphs, resulting in new facets or new result items, or 2) a user
interaction in the UI components demands more data.
4.3 Reactive UI components
The following reactive components were designed to tackle the UI challenges men-
tioned in Section 3. For the sake of terminology, we will be naming entities as categor-
ies and attributes as elements in the following.
8
Common Triple Stores, such as Virtuoso or Fuseki, usually provide a keyword search function-
ality https://jena.apache.org/documentation/query/text-query.html
116
FaRBIE: A Faceted Reactive Browsing Interface for Multi RDF Knowledge Graph
Exploration
Figure 3: (a) Search Box reactive component. A new data source selection produces a
reaction in the categories list with the new entities available in the graph
Search Box: The purpose of this UI component is to provide the user with an in-
terface for the input of the query parameters, and thus the creation of the federated
query. No particular reactivity needs to be incorporated in this component apart from
the common auto-suggestion and auto-completion features well-know for this type of
UI component.
Source Selector & Entity Selector: Allows the user to focus on specific RDF
graphs or entities. The reactivity designed on these components is triggered by a user
interaction. When the user filters a data source, the categories list is updated with the
searchable entities of that RDF graph.
Faceted Bar: Through the use of a real-time-populated Faceted Bar, we attempt to
address the reactivity and visibility issues in terms of views navigation and results (i.e.
data) filtering for a set of RDF knowledge graph datasets. We attempt to achieve this by
combining the Accordion9 user interface element and the List menu patterns into one
unified reactive component, which we call the Faceted Bar. The Faceted Bar provides
users with a layered menu-styled navigational experience, in order to provide the ability
of browsing through entities and thus filtering results as well, in a single component,
which could potentially save time and improve the ergonomics.
For the purposes of our design, we support the implementation of this component
as a left-side-bar or as a top-bar, according to the developer’s preference. The Faceted
Bar also allows to minimize the impact of the screen space trade-off, for the benefit
of saving additional screen space without sacrificing visibility in neither the navigation
area (i.e., the categories and elements filtering) nor the actual screen space required for
the data presentation (i.e., results presentation).
Facet Navigation Menu: The Facet Navigation Menu is a child UI component
of the Faceted Bar UI component. It allows users to semantically navigate within the
elements being obtained as results from the federated search query, through entity cat-
egorization. The idea is to provide menus and sub-menus which allow users to narrow
9
Accordion menu pattern in CSS3.
https://designmodo.com/css3-accordion-menu/
117
FaRBIE: A Faceted Reactive Browsing Interface for Multi RDF Knowledge Graph
Exploration
a
80% 1 0 100% 2 0
b
(1) (2)
(1)
t1 More data arrives (t1+Θ)
Figure 4: (a) Source Box component reacts when a data source query is retrieved suc-
cessfully or with errors. (b) Faceted Bar component reacts to new facet items or cat-
egories.
down the list of results, in order to improve usability by means of offering better group-
ing and increased visibility, while providing a design backbone for future developers at
the same time. Our approach supports default entity categorizations using the types Per-
sons, Organizations, Products, and Documents. For instance, it maps all the elements
obtained as a result of the query which belong to the person category, thus creating a
"menu" container for such attributes. The Facet Navigation Menu UI component holds
two child UI components: Facet Item and Facet Search.
Facet Items: The purpose of the Facet Item UI Component is to serve as the final
display and selection place for each category. Thus, elements such as gender for a cat-
egory of type person, for example, would be placed as checkbox UI components, which
the user could use as selection to further narrow the results view of the query.
Facet Search: The purpose of the Facet Search UI Component is to serve as a
means of a refined search among the obtained elements throughout each category. By
using such component, we provide users also with the ability of refining the field of
options available for navigation of the elements.
Sources Infobox: The purpose of this UI component is to provide a toolbox in-
terface where users can have an overview of the status corresponding to the sources
being queried. This component has three elements: 1) successfully retrieved sources,
2) failed sources, and 3) the information button, which pop ups a dialog with detailed
information regarding the status of the sources, response times, as well as error codes
and messages, which clearly indicate the nature of a failure in a queried source.
Results Container: The purpose of this UI component is to encapsulate the main
UI components related to the results of the query. This includes the screen area allocated
for displaying the list or map of results belonging to a query, in which each result of
the query is then displayed in a child ResultsItem UI component. At the same time, the
children ResultsItem, Source Box, Views Bar and Settings Bar can also be found under
118
FaRBIE: A Faceted Reactive Browsing Interface for Multi RDF Knowledge Graph
Exploration
a b
t1 More data arrives (t1+Θ)
Figure 5: (a) Results Container component reacts when more results are found in the
RDF graphs; additionally, it selects the best view according to the semantics of the
results. (b) View Bar component reacts after analyzing the results, e.g., the map view
is enabled when geo-data is found in the search results.
this UI component. The reactivity designed on this component is triggered whenever
more data is coming from the RDF graphs, e.g., new results are appended to the list.
Results Item: The purpose of this UI component is to provide a unique container
each of the results being obtained from the query will be mapped to, in order to be later
displayed in the Results Container component. Thus, each Results Item UI component
will be shown or hidden from the Results Container, depending on the navigational
input obtained from the Faceted Bar component. The reactivity designed on this com-
ponent is triggered by the semantics of the entities contained in the data, e.g., for a
person, it may be more relevant to show demographic information, but for an organiz-
ation, its location information might be more relevant. This is achieved by specialized
views in the UI component structure hierarchy.
Views Bar: The purpose of this UI component is to provide the users with the
possibility of switching the display of the results through different view modes. The
reactivity designed on this component is triggered by more data coming from the RDF
graphs. After analyzing the results, an appropriate view is automatically enabled. FaR-
BIE supports the following common view modes:
– List mode: This is the default view mode. It displays the results in a list.
– Table mode: It displays the results in a table if a comparison of the entity attributes
is of interest.
– Map mode: It displays the results with location information on a map if geodata is
provided.
– Graph mode: It uses a graph visualization to display the relationships between the
results if a sufficient number of links is found.
Settings Bar: The purpose of this UI component is to provide the user with the
possibility of accessing additional options, such as entering credentials to obtain log-
in tokens, importing a previously saved configuration file to use as input parameter
for the federated search, or selecting the displayed language. Other options could be
implemented and supported by future researchers and developers in order to adapt to
their needs. No particular reactivity has been incorporated into this component.
119
FaRBIE: A Faceted Reactive Browsing Interface for Multi RDF Knowledge Graph
Exploration
4.4 Proof-of-concept
To validate our approach, we have developed a proof-of-concept, based on the intro-
duced use case scenario in the criminal investigation domain. We have configured FaR-
BIE to explore two datasets, namely DBPedia and Linked Leaks. Figure 6 shows the
results using Mauricio Macri as keyword. Two people and one organization were found
matching the keyword, already providing insights to the possible activities and relation
between Mauricio Macri and the Panama Papers scandal, with only a single search.
100% 2 0
(2)
(1)
Figure 6: Proof of Concept. Joined insights from DBpedia and Linked Leaks.
In order to implement FarBIE, we evaluated different frameworks for web user
interfaces as well as web development platforms resulting in the following selection:
– ReactJS10 : A modern javascript library for building web user interfaces. It is a
component-oriented library, and the features of virtual DOM it provides fit perfectly
the requirements of modifying the user interface dynamically when new data is sent
from the server to the user interface. In the work of Khalili et al. [6], it is used as the
core technology to provide a reusable set of user interface elements to build Linked
Data applications.
– Web Socket: A protocol providing full-duplex communication channels over a
single TCP connection. It is an ideal protocol to realize the communication of the
reactive user interface with the backend system. In our case, a web socket is opened
between the LogicKeeper component and the federated search engine. The data is
continuously pushed from the server to the client.
– Play Framework11 : Is a high velocity web framework for Java and Scala. Many
web frameworks, such as Grails, Tomcat, Spring, PHP, or Rails, use threaded serv-
ers. A threaded server assigns one thread per request and uses blocking I/O. The
play framework is based on an event server (Netty). It assigns one thread/process
10
https://facebook.github.io/react/
11
https://www.playframework.com
120
FaRBIE: A Faceted Reactive Browsing Interface for Multi RDF Knowledge Graph
Exploration
per CPU core and uses non-blocking I/O. Threaded vs. event matters in a reactive
user interface, as the engine spends most of the time waiting for query results.
The proof-of-concept interface and the intial source code is available as an open
source project.12 FaRBIE is empowered with flexible user interface components that re-
act to new data coming from the server. The user is able to explore the portion of search
results as soon as they are retrieved from the datasets in the RDF graph federation.
Instant filtering is possible without waiting for the complete set of results.
5 Conclusions
We presented FaRBIE, a reactive faceted browsing UI to explore multiple RDF graphs
from the LOD cloud at a time. FaRBIE follows a reactive user interface approach that
handles the uncertainty imposed by the intrinsic nature of RDF graphs, with the goal
to provide a better user experience. FaRBIE is composed of several reactive UI com-
ponents which react to changes of the semantics, variations in the size of the data, and
the disparities in the response times coming from the different sources, allowing for a
real-time user experience. We argue that the reactive user interface style used in RDF
user interfaces, such as faceted browsing reactivity, is a path for improved overall user
experience and becoming a key factor in bringing Linked Data technologies closer to
non-technical users. For future work, we plan to formalize and generalize our approach,
and perform an empirical evaluation including a user study that compares FaRBIE with
state-of-the-art interfaces. Finally, we will investigate the feasibility of integrating FaR-
BIE with the aforementioned LD-R framework.
Acknowledgments: The presented work has been funded by the German Ministry of
Education and Research (BMBF) in the context of the project LiDaKrA (“Linked-Data-
basierte Kriminalanalyse”, grant no. 13N13627).
References
1. Arenas, M., Grau, B.C., Kharlamov, E., Marciuska, S., Zheleznyakov, D.: Enabling faceted
search over OWL 2 with semfacet. In: Proceedings of the 11th International Workshop on
OWL: Experiences and Directions (OWLED 2014) co-located with 13th International Se-
mantic Web Conference on (ISWC 2014), Riva del Garda, Italy, October 17-18. pp. 121–132.
CEUR-WS (2014)
2. Arenas, M., Grau, B.C., Kharlamov, E., Marciuska, S., Zheleznyakov, D.: Towards semantic
faceted search. In: 23rd International World Wide Web Conference, WWW ’14, Seoul, Re-
public of Korea, April 7-11, Companion Volume. pp. 219–220. ACM (2014)
3. Arenas, M., Grau, B.C., Kharlamov, E., Marciuska, S., Zheleznyakov, D.: Faceted search
over rdf-based knowledge graphs. J. Web Sem. 37-38, 55–74 (2016)
4. Arenas, M., Grau, B.C., Kharlamov, E., Marciuska, S., Zheleznyakov, D., Jiménez-Ruiz,
E.: Semfacet: semantic faceted search over yago. In: 23rd International World Wide Web
Conference, WWW ’14, Seoul, Republic of Korea, April 7-11, Companion Volume. pp. 123–
126. ACM (2014)
12
https://github.com/LiDaKrA/FaRBIE
121
FaRBIE: A Faceted Reactive Browsing Interface for Multi RDF Knowledge Graph
Exploration
5. Ferré, S.: Expressive and scalable query-based faceted search over SPARQL endpoints. In:
The Semantic Web - ISWC 2014 - 13th International Semantic Web Conference, Riva del
Garda, Italy, October 19-23, Part II. Lecture Notes in Computer Science, vol. 8797, pp. 438–
453. Springer (2014)
6. Khalili, A., Loizou, A., van Harmelen, F.: Adaptive linked data-driven web components:
Building flexible and reusable semantic web interfaces. In: 13th International Conference,
ESWC 2016, Heraklion, Crete, Greece, May 29 - June 2, Proceedings. Lecture Notes in
Computer Science, vol. 9678, pp. 677–692. Springer (2016)
7. Nielsen, J.: Usability Engineering. Academic Press, Cambridge, MA, USA (1993)
8. Nielsen, J., Molich, R.: Heuristic evaluation of user interfaces. In: Conference on Human
Factors in Computing Systems, CHI 1990, Seattle, WA, USA, April 1-5. pp. 249–256. ACM
(1990)
9. Stadler, C., Martin, M., Auer, S.: Exploring the web of spatial data with facete. In: 23rd
International World Wide Web Conference, WWW ’14, Seoul, Republic of Korea, April
7-11, Companion Volume. pp. 175–178. ACM (2014)
10. Stolz, A., Hepp, M.: Adaptive faceted search for product comparison on the web of data.
In: Engineering the Web in the Big Data Era - 15th International Conference, ICWE 2015,
Rotterdam, The Netherlands, June 23-26. Lecture Notes in Computer Science, vol. 9114, pp.
420–429. Springer (2015)
122