Keywords

1 Introduction

By using public transport the passenger passes different stages during his journey, from planning the trip up to his arrival at the destination. During these situations the passenger is confronted with a large number of barriers regarding his journey, e.g. missing automation of passenger information as well as insufficient seating capacity. Therefore the passenger needs situational adapted support to master these barriers. This support can be realized by context-aware systems. Context-aware applications respond to changes of actual existing situation and optimize the service or the information automatically in accordance with the user goals. Because of that there is a rising potential of context-aware systems within the public transport.

The context is the initial point for all development activities in context-adaptive systems. Consequently the creation of context models is one of the first steps of development. However, the precondition for modeling is the initial capture of context data. Regarding this the developer teams are primarily faced with two challenges. During the early stages of requirements engineering the developers must have solid knowledge of context and the structure for planning and implementing the data research. Furthermore, the developers have to select appropriate methods for collecting data of potential users. This paper presents an approach for context elicitation of user-centered context-aware systems in public transport to solve the mentioned problems.

2 Background

Before discussing the challenges of context elicitation, a brief introduction is outlining what lies behind requirements engineering of context-aware systems and context of public transport.

2.1 Requirements Engineering of Context-Aware Systems

In general, context can be interpreted as the reification of a real situation [1]. However, from the user-centered point of view the user and his tasks form an inherent part of context [2]. This aspect considers Dey in his general definition by integrating the user in the definition and describing the task in the form of the interaction. “Context is any information that can be used to characterize the situation of an entity. An entity is a person, place, or object that is considered relevant to the interaction between a user and an application, including the user and applications themselves.” [3] Furthermore, context-awareness describes the potential of a system to adapt to the circumstances of a certain users’ situation. Therefore, context-aware systems can provide different functionalities, such as presentation of information and services to a user, performing of services automatically and tagging of context to use information for future retrieval [3]. In contrast to the division of the requirements engineering process into requirements elicitation as well as requirements analysis and modeling [4], this development phase of context-aware systems is enhanced by the requirements monitoring [5]. Here, the possible change of requirements during runtime is taken into account.

However, in this paper, the initial context elicitation is focused. To classify the context elicitation in the requirements engineering, the process by Sitou is presented (Fig. 1).

Fig. 1.
figure 1

Modified representation of the requirements engineering process for context-aware systems by Sitou [19].

The results of the phase scoping deliver the starting point and the limits for the context elicitation in the area of use. Afterwards the context elicitation can be executed by this specification. All information about person, place and any objects which are relevant for the task are collected. As stated by Adelstein et al., elements can be considered relevant if the information is needed for the adaptation of the task as well as the information improves the user’s understanding of the task [6]. The collected data form subsequently the basis for the iterative analysis and the modelling, to generate the user model, the task model and the environmental model.

2.2 Context of Public Transport – A Task-Oriented Perspective

The motivation of a public transport passenger is based on the task in his or her personal agenda. The user needs the public transport to master ways between appointments and tasks of the daily routine. For that reason, the public transport can be considered as a link between single agenda elements [7]. Based on this perspective, the superior goal of “managing a journey” can be derived [8]. When analysing the structure of a journey, it can be divided into a sequence of phases – the so called journey chain. In the view of the Association of German Transport Companies, the journey is broken down into stages and it lists places where the passenger passes during a trip [9]. Hörold et al. divided this journey chain into eight stages, from planning the journey to arriving the desired destination [8] (Fig. 2).

Fig. 2.
figure 2

Journey chain in public transport [16]

3 Challenges for Context Elicitation

The initial points for the presented approach are two essential challenges for the development team during the context elicitation phase: The needed basic understanding of user context as well as the specific requirements for context data elicitation.

3.1 Basic Understanding of User Context

According to Bauer and Spiekermann, the knowledge of the given situation is an essential precondition for identifying relevant context elements [10]. However, the necessary context understanding of the development team and the early execution time of context elicitation are mostly in contrary relation. This problem occurs especially in complex areas of use of the public transport. To specify the needed knowledge of the context for the development team, during the phase of scoping it the supported task of journey chain should already be determined. Thereon the data collection can be executed and subsequently the user, his goal and the resulting tasks as well as the relevant elements of the environment can be characterized.

Because of this, the development team should have a basic understanding and categorization of the public transport context at the beginning of context elicitation.

3.2 The Specific Requirements for Context Data Elicitation

The following enumeration gives an overview of the challenges for context data elicitation:

  • Implicit knowledge of the user

  • Formalization and communication of information

  • Future systems

  • Scope and detail degree of context to be collected

  • Dependencies of context elements

Besides the context understanding, the design of a concept for data acquisition is a central challenge. When developing user-friendly systems, the potential user himself is the centre of data collection. On this account the question arises which methods are suited for acquiring the relevant user data. Context-aware systems are based on explicit and implicit knowledge of the user. In public transport, the implicit knowledge fragments occur e.g. in form of a passengers heavily habituated usage pattern. This kind of information cannot be easily formalized and communicated, because the determinant of an action is done on a subconscious level. Another challenge for the developer as well as the user are visionary systems like innovative assistance systems for mobility impaired passengers. In this case, the future context of an application has to be captured, though the user has only limited experiences of this scenario. On the system side, the developer must not only solve the problem of identifying relevant context elements but also the question of when the context elicitation should be stopped. Schmidt describes this question as non-trivial [11], because the complexity of this decision rises with the required detail degree as well as the number and the heterogeneity of entities, e.g. transport vehicle, driver, stop point, points of interest etc. Besides the scope, the developer has to face several content-related challenges regarding context. For example, in cooperation with potential users he has to define the necessary context elements for the adaption of a task (active) as well as the context element which enhance the users’ understanding of a current task [6]. This poses a particular challenge, because numerous heterogeneous user types with different preferences and information needs exist in public transport [12]. Furthermore, the development team has to find dependencies between identified objects. So objects can have inherent relations [11] or specific values of objects only provide an added value for the application in combination with other objects [6]. In this respect, it is important to remember that users’ context is not only influenced by public transport, but also by the users’ daily routine.

The discussion of the challenge shows, that the characteristic of context aware systems and public transport context may be considered in the concept of context elicitation. Moreover, the consulted users have to be integrated in this decision making.

4 Approach for Context Elicitation in Public Transport

Taking the challenges in consideration the need of an approach occurs in order to overcome difficulties in this domain. Therefore an approach for context elicitation for user-centered context-aware systems in public transport is developed. The first part provides general context taxonomy of the public transport. Furthermore the second part presents a framework for the selection and combination of context elicitation methods.

4.1 Taxonomy of Public Transport Context

The proposed taxonomy can be defined as a categorization of the context information regarding the public transport. The taxonomy is of twofold importance for the interdisciplinary development teams. At first developer teams should enlarge the understanding of the public transport context in order to ensure the precondition of a context elicitation. The second purpose of categorization is the assistance in identifying relevant context elements.

Method of Taxonomy Creation.

The taxonomy is developed within three steps: working model, abstraction and generalization and verification and refinement. In the first step the basic structure of the working model is defined by a combination of top-down-approach and bottom-up-approach. Afterwards the working model is elaborated by a workshop. [cf. 10] During the top-down-process existing context-taxonomies must be checked for suitability regarding public transport by a literature research. On the basis of the selected taxonomy an expert workshop identifies the involved user, the activities and the further relevant context elements with regard to the phase of the journey chain. An analysis of Kolos-Marzuryk identifies these parts as the core of a context model [13]. The extraction of the situational information is done by brainstorming and cord-sorting. To understand the user needs concrete scenarios and persons of the public transport are provided for the participants of the workshop. The participating experts consist of transport company employees and usability engineers with experience in public transport (n = 5). Part of the second step is to abstract and generalize the created working model. Therefore, parts of the public transport taxonomy are derived form the situational information of the scenarios and assigned to the selected taxonomy from the beginning. Afterwards the created taxonomy of public transport is verified and refined iteratively in cooperation with the experts.

Taxonomy of Public Transport Context.

Based on Deys’ definition of context, the primary context types (location, identity, time and activity) [3] are used as the initial point of the taxonomy. However, this classification is not considered to be sufficiently specific in order to describe the situations of public transport in a differentiated manner. Furthermore, the taxonomy should be more focused on the aspect of mobility. Therefore the existing categories are expanded by the categories information, social and environment [14].

Overall, the proposed approach to the description of public transport context includes seven different categories: user, goal and the associated tasks, spatial context, temporal context, environmental context, social context and informational context. Besides these basic categories the taxonomy contains different degrees of abstraction, macro level, micro level and situational level [cf. 10]. The level with the highest degree of abstraction (macro level) is valid for all context-aware application in public transport. On this level the overall context of public transport is subdivided into the relevant categories. The second level (micro level) substantiates each category and provides support for the identification of relevant context elements in specific application environments. Finally, the last (situational level) describes the relevant context elements of a given situation.

The taxonomy of the public transport should be introduced by an example of the journey chain, the phase of the transfer to another vehicle (Table 1).

Table 1. Taxonomy of the public transport

4.2 Framework for Selection and Combination of Context Elicitation Methods

As the second part of the approach a framework for selecting and combining of data acquisition methods is provided. The goal is to support the developer team by selecting the methods for an effective and efficient context elicitation regarding the prevailing conditions.

Structure of the Framework.

Based on the discussion of challenges above, four main sections can be identified influencing the concept of a context elicitation in public transport. These sections are the basic structure of our framework in combination with the methods proposed. The first section, user, consists of aspects relating to the user as a knowledge source. In that we distinguish between the organizational factors and the characteristics of the user which could influence the quality of the collected insights. The second section lists the factor related to the investigated tasks. Moreover, the third section includes the pertinent factors regarding the context. The last section takes account of the investigator and project conditions (Table 2).

Table 2. Framework for selection and combination of context elicitation methods (based on [15, 17, 18]).

The involved methods are an extract from the spectrum of knowledge elicitation and can be grouped into three categories: creative techniques, observation, and inquiries. The selection is focused on the wide distribution and the manageable complexity in order to ensure the applicability.

The framework is intended to find a method for context elicitation in public transport. Because of this a description of the methods is not listed. A detailed description of the methods is, inter alia, discussed in [15, 17, 18].

The suitability evaluation of each method is carried out through a three-step scale: un-recommended, no influence and recommend. Each assessment shows, whether the method is recommended (+) or unrecommended (−) by the respective factor. A neutral factor which has no impact on the method is characterized by no influence (0).

Application of the Framework.

Finally a three-staged procedure for selecting the suitable elicitation method should be introduced:

  1. 1.

    Preselection of methods based on the kind of examined knowledge

  2. 2.

    Analyzing the factors

  3. 3.

    Creation of a method mix

According to Rupp a distinction can be drawn between subconsciously, consciously and unconsciously knowledge. Based on this point of view a tendency regarding the suitability of the method can be concluded. Observing techniques are especially qualified for subconsciously knowledge. In contrast to that the best way to figure out consciously knowledge are questioning techniques. Unconsciously knowledge, e.g. in case of visionary tasks, appears mainly by using creative techniques [15]. During the second step the methods are compared among themselves according to the factors and potential methods are selected. Finally it is possible to compensate appearing weaknesses of the selected methods by combining them with further methods.

5 Conclusion

In this paper a task-oriented procedure is presented for the context elicitation in public transport. Based on the user’s tasks, general context taxonomy of the public transport is developed in the first part. This context categorization pursues the aim to enhance the understanding from the developers of the context, since this knowledge of the user´s context is the prerequisite for the early phase of the requirements engineering. Furthermore, the taxonomy facilitates the identifying of the relevant elements in the context of the public transport. This can be achieved because the tasks of the user limit the relevant scope of the situation. Beyond that the written composition shows the challenges of context elicitation for context-aware systems in the area of public transports. Based on these requirements, there will be developed a framework for the selection of an effective and efficient elicitation method. Finally, the usage limits of the framework should be discussed. In our understanding the taxonomy is only a tool for eliciting and structuring of the context. On this account the given taxonomy should be verified for validity in all developments and if necessary, depending on the aim, be adapted to the application. Regarding the method selection, there is an agreement with Rupp, who points out that the ideal method cannot be defined generally [15]. Therefore, the selection of factors have to be enhanced or adapted according to the prevalent project terms even for this introduced framework. Besides the factor the methods can also vary. The present selection is oriented on the broad distribution and the simple use. Depending on the project the usage of expansive methods could bring a clear added value.