This position paper argues for adapting the Apps-approach to industrial interactive Linked Data applications (iLD) and presents a design study for a small application that is applicable whenever an information foraging task at hand may be described as carefully examining the neighborhood of a concept. This little App is primarily being developed for industrial device diagnosis tasks. However, it is easily adaptable to other navigation and information processing tasks that have similar information needs. We argue that a collection of individually extremely limited but highly usable applications will eventually have the potential to provide the means for application orchestration as a complement to industrial Linked Data Clouds and service orchestration. We present the concept of a generalizable mobile user interface that hides the complexity of the underlying Linked Data Network from the user.
INTRODUCTION Due to ever shorter release-cycles and growing pressure to innovate, independent small and medium enterprises are entering short-term cooperations to be able to develop and produce marketable products. In particular in fast moving markets, the concept of virtual companies (a loosely coordinated set of complementing companies that temporarily work together) promises a competitive advantage over larger enterprises.
While several approaches to meet this challenge have been attempted in the past, there is still no single accepted standard for controlling and exchanging data over the companies’ boundaries of trust. This is even true for monolithic companies if one takes the complete information life cycle of a plant into account. Conventional paths to unify the “standard” ERP systems, for instance after a merger, are usually very time-consuming. In the process and manufacturing industries, several approaches to come up with a single unified world-model (for instance STEP with its domain-specific APIs [5] or more recently ISO 15926 [3]) have been tried. Yet, all these world model approaches have turned out to be too complex and too unspecific to meet the individual requirements without customization. So far, the most promising approaches in that area seem to be metamodels (e.g. CAEX – Computer Aided Engineering Exchange) which provide enough formalism to allow transformations between particular implementations of different partners during the life cycle of a product or a production plant.
If the higher requirements for trusted and secure information
exchange are neglected for a moment, the need to
coordinate large distributed information islands is quite similar to
that of the public semantic web. Efforts to organize the
immense amount of unstructured or semi-structured data
available on the web face the same challenges. However, there
are organizations such as the Linking Open Data movement
(LOD, [6]) that strive to give more structure to the data cloud
through Linked Data. The basic idea of Linked Data is to
augment the large amount of publicly available data by
exposing it through a computer interpretable structure, based
on web technologies. It uses URIs as names for things so
that people (and machines) can look those names [
In the following sections we will give a survey on recent interfaces to Linked Data Clouds. We then propose to identify generalized information handling tasks that allow us to create a set of interactive Linked Data Applications (iLDApps) which have the potential to be highly usable. As an example of this approach the analysis of a shop floor maintenance scenario is given. The cognitive process when diagnosing a problem during a maintenance course involves the evaluation of attributes of the device under inspection by comparing it to observations in its neighborhood. This neighborhood can be represented by specific relations of a directed graph constructed from Linked Data. A generalizable mobile user interface concept that hides the complexity of the underlying Linked Data Network from the user has been developed.
RELATED WORK
Linked Data Browsers
Displaying and navigating RDF data in a user-friendly way
is a problem addressed by various types of applications
using different representation paradigms. According to [
Recognizing the general need for presenting RDF content
to users and wanting to promote the exchange of
presentation knowledge [
Mobile Linked Data Browsers
All of the mentioned approaches call for large displays and
are designed with the WIMP-Metaphor in mind. Therefore,
these concepts may not be transfered directly to mobile
devices. It is not surprising that as of today one can find only a
small number of mobile Linked Data applications.
A rather prominent instance is DBpedia Mobile [
MOBILE MAINTENANCE SCENARIO Currently available Linked Data Interfaces address very high level generic tasks such as searching, visualization and navigation. While these paradigms are highly generic for industrial applications we propose some adaptation.
We suggest to find and implement interaction metaphors that are closer to the task. This means on the one hand to rely on asserted task specific invariants of the linked data cloud, on the other hand to allow for changing relations and sets of attributes.
A prerequiste for the delevopment and orchestration of this kind of interaction metaphors is a sound task analysis. Before we go into details of a single metaphor concept for diagnostic tasks in operation and maintenance, we therefore describe the environment and task at hand.
Environment
As industrial manufacturing facilities have become more and
more autonomous, management and monitoring have
become increasingly centralized in large control centers, which
are not necessarily located close to the facilities themselves
[
Maintenance and Diagnosis Tasks A requirements-analysis in cooperation with industrial partners has lead to the hierarchical task model shown in figure 1. The goal perform maintenance circuit is comprised of two operations and one sub-goal. The sub-goal complete task is comprised of six operations. It must be noted that the model only reflects the mobile part of the fieldwork. Additional parts like negotiating access to the plant control room, getting dressed or preparing equipment have not been considered.
0
In our use case we assume for subtask 3.3 to be a diagnosis task at the checkpoint. The process value given by a measurement device has to be inspected and evaluated. A common strategy is to compare the reading of the device under inspection with other measurement devices that reside upand downstream. To identify these reference devices, one follows all the connections along intermediate instruments in a piping and instrumentation diagram (P&ID - a graphical 2D representation of the piping and process flow together with devices). When a dependent measurement device that informs about the particular process value is found, it can be systematically assessed.
Linked Data Model With todays computer-aided engineering systems, all of the information necessary to generate a Linked Data cloud is potentially available. Information on devices, equipment, pipes, materials, as well as their relationships can easily be retrieved. Figure 2 illustrates the case of a measurement device P0001 that indicates pressure at Tank010. Pump010 delivers the medium from the tank into Stream010. The pressure in that stream (after the pump) is observed by indicator P0002.
Orchestration of Applications It has been observed that mobile users tend to individualize their devices with a personal selection of applications addressing their needs for mobile information. Drawing from <#P0001> rdf:typeOf <#meas_device>. <#P0002> rdf:typeOf <#meas_device>. <#Stream010> rdf:typeOf <#stream>. <#Tank010> rdf:typeOf <#storage>. <#Pump010> rdf:typeOf <#equipment>. ... <#P0001> svg:isConnectedTo <#Tank010>. <#Tank10> svg:isConnectedTo <#Valve010>. <#Pump010> svg:isConnectedTo <#Stream010>. <#P0002> svg:isConnectedTo <#Stream010>. ... this phenomenon, our approach builds on a set of specialized small applications focusing on an informational need that is common on the shop flor.
The orchestration of simple single-purpose applications is also useful in the industrial sector, e.g. as a front end to available services. These services implement the needed operations on the Linked-Data-Cloud, such as SPARQL queries. Depending on the context (various platforms, various stakeholders), well-adapted applications can be provided for the services. Model-driven development methods can support and accelerate the development of such applications. The orchestration itself, i.e. collection, arrangement and interaction of the individual applications, can also be effectively supported by automated model-driven methods.
The portfolio of applications can be expanded and updated, without the need to reprogram the mobile front-ends. By simple adaptation of the orchestration, the mobile front-ends can efficiently be adapted to changing workflows and user roles, again without having to reprogram the applications. Thus, a previously unknown degree of flexibility, innovation and usability is achieved. At the same time, the applications have the potential to become more consistent, require less maintenance, be safer and more reliable.
Neighborhood Browser In industrial contexts various informational needs may arise, e.g. during commissioning and loop checks. Several of these needs may be addressed by the proposed neighborhood browser. Off course the details – i.e. which particular “things” are of interest – are quite variable. However, to the authors understanding, this is exactly the strength of the loose coupling in a Linked Data world.
In order to evaluate a particular measurement device,
maintenance personal needs to obtain information about the
connected devices. Based on the findings in [
It has proved to be advisable to provide the complete tags
(i.e. unique labels) of the devices and display process values
as well as engineering units. Maintenance engineers need to
be able to grasp all necessary information on the type and
state (open, close, on, off, powerless) of the intermediate
equipment at first glance. Neighboring devices can be
chosen as the new “center of the universe” to be displayed in the
center of the screen. The application can obtain the identifier
of the device to be investigated e.g. by RFID or Data Matrix
code [
For different use cases different interface patterns must
complement the Neighborhood Browser. Therefore, to complete
the application orchestration, it is an ongoing work to
identify further mobile interaction patterns. Timelines and
calendars immediately come to mind – they could be presented in
a linear form (as in Tabulator) or in form of tubes or worms
(as in DySoN) [
We are currently investigating on further iLD-App patterns that address other aspects of mobile day-to-day work in process and manufacturing industries. For further promotion in industries an extensive evaluation of user experience with this approach will be conducted.
Generalization The particular presentation and interaction scheme, as presented above, may be easily generalized to other information foraging tasks that focus on the close neighborhood. Just to illustrate the basic idea, figure 4 shows the SPARQL-query that would be necessary to gain information about closely related programming languages from DBpedia. In combination with a set of concept specific sorting, aggregation and presentation ontologies and a basic set of interactionmetaphors the proposed concept may be easily converted to a programming language browser.
PREFIX dbpprop: <http://dbpedia.org/property/> SELECT ?left ?leftname ?right ?rightname WHERE { { <http://dbpedia.org/resource/XML> dbpprop:extendedTo ?left .
OPTIONAL { ?left dbpprop:name ?leftname } } UNION { ?right dbpprop:extendedTo <http://dbpedia.org/resource/XML> .
OPTIONAL { ?right dbpprop:name ?rightname } }
} CONCLUSION Orchestrated iLD-Apps can be a usable approach to meeting the challenges of data exchange in virtual companies. Taskcentric applications, based on Linked Data, help to obviate cumbersome interaction of maintenance engineers with devices during diagnosis-tasks. The proposed Neighborhood Browser can be generalized to similar use cases. The best practice of how to adapt the proposed approach to address schemes, diagrams and plans which are of utmost importance to industrial engineering has not yet been conclusively found. Nevertheless, the authors are convinced that the Linked Data approach, utilizing model-driven development paradigms, is well-suited for the rapid development of a highly usable ecosystem of specialized but generalizable applications.
//www4.wiwiss.fu-berlin.de/bizer/ng4j/disco/.
Integration of life-cycle data for process plants including oil and gas production facilities. ISO 15926. 4. Interactive relationship discovery in RDF datasets.
http://relfinder.dbpedia.org/. 5. Standard for the exchange of product model data. ISO 10303.
http://esw.w3.org/SweoIG/TaskForces/ CommunityProjects/LinkingOpenData.