Recently, semantic technologies improved the information access to massive amounts of data in various manners. Also industrial domains bene t from semantic approaches. Information now can be accessed more precisely and is directly linked with related data. However, the actual support for humans to access suitable content at time is still in its infancy. In this paper, we motivate a number of requirements for extending current state-of-the-art information systems in order to support human-aware information access. The requirements are based on the experiences we made during the introduction of industrial information systems over the last years.
We motivate the application of semantic technologies and a ective computing in the domain of Technical Service. The primary task of Technical Service is the troubleshooting and maintenance of advanced machinery. For instance, Technical Service has to identify the reason for a malfunction of a machine and in consequence has to replace and adjust faulty components. With the increasing complexity of machinery the Technical Service becomes more and more challenging even for well-trained service technicians. Technical documentation provides the service-related information to support technicians during their work. Having complex machinery, however, the information typically covers some thousand pages only for a single machine. Consequently nding relevant information units in a problem situation is di cult, time-consuming, and often not successful. Many machinery companies report, that their globally distributed service technicians are often not able to nd helpful information during their operations.
Recently, semantic technologies advanced information systems to improve
the search and navigation in the information data bases [
In this position paper, we discuss the requirements of the domain Technical Service in Section 2. We propose a conceptual model approaching these requirements in Section 3. We conclude the paper in Section 4. 2
Semantic technologies improve the accessibility of resources in information systems dramatically. In comparison the standard information systems with textbased queries and tags, semantic information systems are able to precisely point to resources based on concepts stated in the query. Further, they are able to rewrite queries so that results include semantically related or synonym resources.
Nevertheless, a number of requirements exist in Technical Service, which should be taken into account to simplify and re ne the information access for the user. We claim, that current systems miss di erent aspects of awareness that in summary complicate the access to information. An overview of the aspects is depicted in Figure 1.
Context-awareness: - Product under work - Problem context - Location of work - Available equipment - Already consumed knowledge
Car ier htp:/www.domain.com 12:00PM ServiceMate
Go gle
- User persona - Already researched & consumed knowledge User-awareness: - Already consumed knowledge - Emotional state - Affect detection - Emotive UI - Body sensors - Multi-modal input 1. Context-awareness: For large corpora, even semantic search yields large amounts of information resources and the selection of the most appropriate resource rarely can be done automatically. With the knowledge of the current user context, a context-aware system can lter the query results to the most appropriate ones. For instance, when looking for the replacement of a component, it is not necessary to also o er the operation manual for this component. 2. User-awareness: Stating a semantic query is not necessarily more userfriendly than a standard text query. Users have frequently problems formulating appropriate queries. With growing frustration of not nding an appropriate information resource, the users are again turning away from the documentation system missing important information for their work. A user-aware system will monitor the user's satisfaction during the information research and consumption. It will propose alternative ways of communication when the user's satisfaction is downgrading. 3. Competence-awareness: Technical information is typically written for well-trained technicians. In the context of world-wide service operations the technicians may not be well-trained in some regions. Then, the information resources are simply not useful because they miss important detail for less competent technicians. A competence-aware system will preferably present information with an appropriate level of detail to the user.
In the following section, we introduce a conceptual model for the realization of the requirements stated above. 3
We develop a conceptual model relating information resources from a semantic information model with concepts representing context-awareness, user-awareness, and competence-awareness. The semantic information model is tted to the needs of information management in the domain of Technical Service and has been successfully used in a number of industrial projects.
The shown model comprises task-speci c ontologies and it will be used as the
blueprint for implementing smart service systems in the future. For the
implementation and visualization of the ontologies we used the open-source version of
the semantic wiki KnowWE [
Figure 2 shows the general concept of information resources, i.e., entities.
We use the class prov:Entity from PROV-O [
MimeType has_annotation skos:Concept inverse skos:narrower are used with corresponding sub-properties to build a exible hierarchy structure.
The basic ontology from above is extended to model the general artifacts of the domain Technical Service. An actual application{the entities of a concrete manufacture{will align to this upper ontology.
In Figure 3 we see the class sm:Concept sub-classing from the general class skos:Concept (the pre x sm stands for service model ). The class sm:Concept represents general entities the can build of elements with the same type. For example, we usually de ne a hierarchy of sm:Product instances, where each products consists of hierarchically structured sm:Component instances. Usually, a component ful lls a speci c sm:Function. Final instances of sm:Component can be referenced to sm:Part instances. Please notice, that a real-world ontology additionally describes more detailed types of components, for example, classes representing electrics/hydraulics, located components, and virtual components.
In Figure 4 we see an example with concrete instances (pre x ex is used for example instances). When describing the component climatic system, we see that this system (among others) consists of parts like AC pump and a climate control module. Speci c components ful ll speci c functions of the machine. In the shown example, the component climatic system is related with the function cooling and the function heating. Having represented the product world of the application, we are able to use the instances as annotation values. In the example, the concrete book ex:book4711 consists of a chapter for replacing the climatic system (ex:acReplacement) and a chapter showing the wiring diagram (ex:acWiring). The rst chapter is assigned to the pump of the climatic system, i.e., how to replace the referred pump. The second chapter depicts the wiring of the corresponding climatic control module.
In an information system, the query \give me information about the component ex:climaticSystem" will nd both resources ex:acReplacement and ex:acWiring, since the reasoning will show that the resources are assigned to parts that are included in the climatic system. Also, when looking for electric sm:Concept prov:Entity a
subClassOf refers_function sm:Function information for the function ex:cooling, the system will yield the information resource ex:acWiring, since the semantic reasoning will derive this resource as an appropriate (because close) match.
In the following, we brie y sketch how the ontology can be extended in order to ful ll the requirements of an a ective information system. All these attributes can be used to further reduce the number of information resources shown as query results but also should be used to present the appropriate user interface depending on the context and state of the user. 3.2
In summary, the context of a technician comprises: { Product under work, i.e., the actually considered product the technician is working with. { Problem context describing the activity of the technician, e.g., assembly vs. diagnosis.
For larger problem contexts the current working step is also an interesting attribute of the context. { Location of work, e.g., workshop vs. eld. { Available equipment, e.g., which level of tools/measurement devices are available { Already consumed knowledge, i.e.. which information resources the technician already used during the current problem context.
Context-awareness should be primarily used to further lter the query
results [
The user-awareness can be attributed to the following information: { Consumed knowledge identifying which information was already seen by the technician. { Emotional state of the user showing the satisfaction with the current situation. The situation can be possibly derived from the software functions or problem context. The user's satisfaction may be derived by a ect detection algorithms and/or using body sensors.
Knowing the state of the user the tool interface should adapt accordingly. Adaptation of the software can be implemented by multi-modal user interfaces. For example, after realizing that a speech recognition did not succeeded for a couple of tries, the system should switch to alternative input interfaces. 3.4
{ User persona means distinguishing expert vs. apprentice. A user persona is identi ed in order to present an appropriate interface and information granularity to the user. { Already researched and consumed knowledge can be used to derive the competence level/needed information granularity of the current user.
Competence-awareness should be used to select the appropriate information granularity for a speci c query. For instance, the system can decide to immediately show the disassembly of a speci c component (expert level) or to rst explain the location and operation of the component, before presenting its disassembly (beginners level). The disassembly itself could also distinguish the user's competence: Experts may only need a summary of steps with service-relevant facts (bolt strengths, pressures, etc.), whereas for beginners the systems shows a step-by-step disassembly instruction possible supported by 3D animations for better identi cation and orientation of components.
In this position paper we introduced a number of requirements of information systems posed in Technical Service. The requirements are based on a 10-years experience of establishing and improving information systems in this eld. We claim, that current state-of-the-art information systems need a ective extensions in order to cope with this requirements. In the previous sections we sketched possible ways of extensions. Although the presented approach is by far not exhaustive, we see a eld of exciting research questions with a practical motivation in industry.