Abstract

The thesis investigates the value that mobile computing devices add in seeking, accessing and retrieving material from hybrid information collections. The ability to maintain access to electronic content and services while moving and exploring the available physical information domain is expected to enhance the seeking experience. Beginning with a summary literature review, I suggest a service architecture, an evaluation model and a plan to explore the added-value of a new seeking approach. I conclude with planned analysis procedures and future research directions.

1. RESEARCH AREA

The portability of mobile computing devices such as Personal Digital Assistants (PDAs) and their capability of carrying around information to be available for reference or advice anytime/anywhere was quickly acknowledged by field practitioners [1], upon their appearance in the mid 1990. Nowadays, these devices have evolved to always-around, powerful gadgets capable of telephony services and functionalities to exchange data with local or remote computers and information management systems. They are also equipped with long-life batteries and high resolution screens that allow on-screen text reading and multimedia content presentation. Several researchers and firms around the world explore their applications as tourist/city/museum guides, mobile learning tools or even recommendation wizards, and mobile commerce terminals [2-5]. In particular, in the field of Information Science, many libraries expand their collections with information material in electronic form and develop services to support their PDA/smartphone patrons with innovative functionalities and full-text-content, properly formatted for the small form of these devices [6-7]. The ability to freely navigate and browse the stacks of books while maintaining access to the electronic resources and services brings closer the two information domains (physical and electronic), thus creating a new domain which is sometimes called hybrid. Especially in the field of Librarianship & Information Science, several initiatives and projects were started to support research for the Hybrid Library of the future [8].

In many cases, information to be obtained from multiple data items requires significant processing, e.g., development of mental models, exploration, etc., for people to gain insights to information underlying the data, which is where most computer-centric techniques fail. The human mind outperforms any currently known information retrieval system in extracting information and obtaining insights and therefore it is crucial to keep people in the center of the information-extraction procedure when dealing with environments of big and diverse information collections as in hybrid domains. Dr. R. Jain [9] calls these environments experiential, because the user is allowed to explore, experience and directly interact with the available data, whereas Dr. G. Marchionini coined the term Human-Computer Information Retrieval (HCIR) to describe information retrieval techniques that bring human intelligence into the search process [10].

Due to inherent portability of the mobile computing devices, library patrons can use them to seamlessly explore both physical and electronic resources simultaneously, without moving from the physical domain to the electronic (or vice versa). Until recent years, precision and recall were the dominant criteria to assess information retrieval processes and tools [11]. However, seamless exploration of the available material enhances serendipity and diversity, which are increasingly seen as important for the assessment of efficiency and effectiveness of seeking procedures [12-13]. The extent to which mobile computing devices can add value to the currently available services for seeking, accessing and retrieving information in hybrid environments, and the identification of the factors that determine the acceptance and usage of such services, along with their interactions, are the topics of my thesis.

2. AIMS, OBJECTIVES and METHODOLOGY

The aim of my research is to examine whether mobility and seamless navigation in hybrid information collections adds value to the currently available searching functionalities offered by the institutions that hold such material. In order to explore the human interaction with the unified information space, we need to describe, develop and evaluate a prototype architecture to support the new seeking approach. Therefore, my objectives are:

• gather requirements for the prototype: start by defining a set of prototype requirements, determine its functionalities, identify the user-system interaction points, and describe the interfaces. In addition, we need to determine the evaluation criteria, metrics, and tools to assess the quality, usage and intention for future use of the new service.
• construct the prototype: this objective involves the implementation of the prototype, i.e., determine and install the appropriate technical equipment, choose appropriate technical solutions to integrate prototype with existing infrastructures, select appropriate development tools, create and implement necessary software for the data exchange among distributed computing resources, e.g., protocols and scripts.
• evaluate the new service: this objective involves the evaluation of the prototype. It involves the design of an evaluation plan that describes the evaluation model (relationships among criteria), the research hypotheses, the data collection method and instruments, the experimental procedure, the appropriate data analysis method, and finally the presentation of results and findings of the evaluation.

Our methodology's first step is a survey of the literature in the fields of mobile computing and information science. This survey provides information on the current trends and approaches in these fields and some guidelines to follow, difficulties to expect and appropriate research methods to use. Next, a focus group conducted with experts from the fields of Information and Computer Sciences can determine the design requirements of the new service, its functionalities to implement and the architecture model to use. The next step involves the installation of the prototype system and its integration with the available infrastructure of an actual library that has a big, hybrid collection. Beforehand, we plan on having some pilot tests to identify any problems related to the proper operation of the service, the measuring instruments and the evaluation procedure itself. Finally, we plan on having a field experiment to evaluate the new service and prototype it in a real-life situation, with regular users of the hybrid collection.

Our expectations are that the analysis of data resulting from the service's evaluation will reveal its benefits and weaknesses, the extent to which mobility helps the users in experiencing hybrid collections and gain insight from the available data, and will also provide some understanding about the users' interaction with the hybrid collection. Finally, feedback from the users will be used to start a redesign cycle of the prototype system, which will enhance it and generalize its architecture, so that it can be used in other rich-information, hybrid environments, such as museums, memory institutions and historical centers of popular cities in Greece.

3. WORK DONE SO FAR

In the following paragraphs I will outline the steps being undertaken and point the progress to date, according to the methodology described above. Starting with the procedure of extracting the requirements and specifications of the development of a prototype system to support the new seeking approach, I move to the description of the architecture adopted and finally I describe the evaluation plan to assess the new service's effect on its users.

3.1 Creating the new service

Literature review: starting with a survey in literature of mobile applications in rich information environments (libraries, museums, etc) we found out that most of the applications could be divided in 3 categories: (a) reference services, (b) searching tools and aids, and (c) synchronous/asynchronous communication. This is consistent with information seeking behaviors, where users usually combine and interchange browsing, searching and experts-asking methods in order to retrieve the information they need. Typical examples of these clustered applications are: access to locally/remotely stored, (un)structured electronic data, such as the WWW and digital libraries (DLs); access to on-line services and use of other data retrieval tools, e.g., search engines, directories, thesauri, indexes, maps and code scanners; and use of communication tools either in real or past time to ask for the help of an expert or contact a colleague anytime/anywhere.

The literature review also revealed that the development of services to be used in a mobile context is quite different from that of a controlled environment [14]. In general, mobile elements like PDAs are resource-poor relative to desktop computers; mobility is inherently hazardous since mobile data terminals are more vulnerable to loss or damage; wireless connectivity may be highly variable in performance and reliability; all mobile elements rely on a finite energy source; and task interruptions occur more frequently since the users are more likely to be affected by the changing context that surrounds them.

Another aspect to take into account for the development phase of the new service is the profile variability of the users for whom the service is intended to be provided. We needed to take into account the users' profile and their information needs. For example, not all library patrons have the same information needs; pre-graduate students usually need to find quickly a few good documents on a particular topic, whereas most post-graduate students and researchers need to find all relevant documents for a thorough understanding on a subject. There are also some users with low expertise in computer handling and others quite efficient in using computer technologies.

To deal with all these aspects we decided to start with a focus group of librarians and ICT experts, which would be an appropriate foundation to build a development plan for the new service.

Focus group: we conducted a group discussion with three objectives: (a) obtain knowledge about our intended audience, i.e., their usual seeking strategies, their information needs and their computing experience, (b) identify the technical difficulties to be expected in the development of our prototype system that will be used to support the new service, and (c) determine which mobile applications are more likely to be useful when exploring hybrid information environments and which are the key tasks to be supported. The group consisted of 5 librarians from academic libraries and 4 computer experts with prior knowledge in mobile computing applications. A moderator was controlling the discussion and two observers were taking notes. The discussion was video recorded so that it could be later reviewed. A limited summary of findings follows below.

• Academic libraries have a wide audience since they serve students and research personnel from several academic departments. Thus a variety of user profiles in terms of skills, information needs and seeking experience should be taken into account during the development process. In addition, the prototype system should be at least as effective and easy to use as the current one for all groups of its intended users.
• Many of the constraints of a mobile working context are easily handled as long as the user remains indoors. For example wireless connectivity can be reliable and fast; the user can benefit from nearby resources like bigger screens and keyboard; during idle times the device can be recharged. Code scanning methods are attractive, but the designer has to deal with the problem of scanning speed and variety of implementation codes and protocols (barcodes, quick response codes, Radio Frequency Identification [RFID] tags, etc).
• Key tasks to be supported where anywhere/anytime access to the currently available searching tools and to content used for reference, (a)synchronous communication, and easy data transfer from the mobile device to another computer. However, librarians were skeptical about personalized services due to security vulnerabilities and personal data violation. Therefore, they were not considered a first priority for implementation.

Apart from the discussion records, in order to have a priority list of the functionalities and applications to be implemented, we asked the participants to rate a set of proposed functionalities on 5-point Likert scales so that these could be ranked. In the following paragraph we present which of these functionalities were adopted, according to usefulness contribution and technical feasibility criteria, and how these can be combined to form a typical usage scenario.

Service functionalities and usage scenario: the users' scores for the functionalities of the prototype system along with their average (AVG) values and standard deviations (SD) are: (a) wireless access to book catalog and full-text electronic databases (AVG=4.63, SD=.52), (b) a floor plan indicating a book's location in the stacks (AVG=4.25, SD=.70), (c) communication with other people over email or instant messaging (AVG=4.13, SD=.99), (d) ability to download, disseminate, share and synchronize information content with desktop/laptop computers (AVG=4.38, SD=.52), and (e) taking a quick note or reminder on an electronic notepad (AVG=4.0, SD=.75). These functionalities can be combined to support the user while exploring a hybrid data collection as in the scenario described bellow.

A student submits a query to the mobile, on-line public access catalog of the library. In the results list, she spots the desired book and other related print works. With the stylus she taps on the record of her choice and retrieves its metadata to check its availability and location. While walking to the stacks she activates her instant messaging account. Having found the book, she takes a memo of other related books and sends a short message to the on-line librarian, asking to inform her of due dates of previously borrowed books. Without needing to leave the stacks, she searches the library's electronic collections for relevant records. She decides to download a couple of them and move them to a desktop computer where she can have a closer look. In addition, she sends their metadata (including a download link) to a co-worker by email.

3.2 Service Architecture

The client-server model: we decided to use a client-server architecture so that any computation required for the retrieval process could be carried on powerful remote servers, whereas the mobile devices would be used for data presentation. As shown in Fig. 1a we chose a 3-tier architecture model in which the user interface, functional process logic, computer data storage and data access are developed and maintained as independent modules. In addition, this model allowed us to easily integrate the new modules of our prototype system to an existing infrastructure.

All data presented to the user is in the form of a small web page, so that the new service can be available to any device capable of web browsing. For example, a user interface on the mobile device is used to submit a query (presentation tier), which is then sent to and interpreted by a web server (logic tier). There, a proper-syntax query to the available databases is formed and sent to storage machines (data tier) in order to retrieve any matching records. The list of these records is sent back to the web server, where it is properly formed and finally sent back over the wireless network to the mobile device, to be presented to the user.

(a)


(b)                                 (c)

Figure 1. (a) The prototype architecture, (b) the search interface, (c) the evaluation model.

Combining information from multiple sources: for seamless exploration of the hybrid collection it was important to minimize the searching effort and provide easy access to multiple searching targets. Therefore, we used a federated search engine, called dbWiz [15]. This open-source engine can search multiple sources by submitting a query once and return matching records from selected sources, such as library catalogs, citation/full-text article databases, web sources (Google, Amazon, etc), and SQL databases. In addition, new drivers can be written in PERL programming language so that additional sources, e.g. XML files, can be parsed. The federated search engine was implemented in the logic tier of our prototype system.

The mobile user interface: as revealed from the focus group, our intended audience would be highly variable in terms of computing experience and searching skills. Therefore, the human-computer interfaces should be easy and simple to use without distracting the users' attention from their surrounding environment, letting them explore and experience it; we also had to keep mental effort and memory load as low as possible since frequent interactions are expected in a mobile working context. In addition, we had to design the interfaces for both novice and experienced users. To meet all these requirements, we studied and followed the Human-Computer Interaction guidelines from firms with large experience in mobile applications, such as Apple and Microsoft [16-17]. All interfaces we designed were in the form of web pages, which involved learning and handling web programming (HTML, CSS, XML, PHP, etc). Another issue we had to deal with were the Greek characters that should be properly retrieved from the databases, transferred and displayed to the user. Fig. 1b shows the dbWiz form that is used to submit a query.

3.3 Service Evaluation

Criteria and evaluation model: the extent to which the new service adds value to the seeking and exploring experience when searching in hybrid collections is measured by recording the users' behavioral intention (BI) (willingness) to use the new service. We are also interested in factors that directly or indirectly affect BI, such as performance expectancy (PE) and effort expectancy (EE), as well as the interactions between them. These are often called second order factors (or constructs) since they can be inferred from other evaluation criteria (first order) such as attitude towards using technology (ATUT), perceived usefulness (PU), job fit (JF) of the new service, relative advantage (RA) in contrast to currently available alternatives, and outcome expectations (OE) from using the new service.

Table 1 shows the metrics used to assess the evaluation criteria. These are a set of self-reported, subjective user perceptions regarding how the users found the new service. They are commonly used in Davis' Technology Acceptance Model (TAM) [18] and Goodhue's Task Technology Fit (TTF) [19]. Our evaluation model, shown in Fig. 1c, is based on UTAUT which is an integration of TAM, TTF and 6 more models [20].

Data collection instruments: the users' subjective perceptions of the criteria mentioned above, were captured on multi-item scale questionnaires. These scales were adapted from the literature [20], since they were well studied and validated. Prior to using them, we pilot-tested the questionnaires with a sample of 40 students. Cronbach's α reliability scores for multi-item scales ranged from .75 to .95 which are above the generally acceptable level of .70. In addition, we slightly modified a QUIS (Questionnaire for User Interaction Satisfaction) [21], which includes items to record users' subjective satisfaction with specific aspects of human-computer interface, such as general interaction attitudes, screen and presentation, text input, language and system response, learnability, and service capabilities. These questionnaires can be found at http://dlib.ionio.gr/hls/

Adopted research and analysis method: since there is little knowledge of the exploration of hybrid information collections with mobile devices, we chose to conduct an exploratory study, i.e., define possible relationships (effects) among evaluation criteria as shown in Fig. 1c and then use multivariate techniques to estimate a relationship. The experiment evaluation was planed to take place in Patras University Library (Greece), where students and researchers would be invited to use the new service as described in the usage scenario, in subsection 3.1, and provide feedback with comments and valuable measurements to be analyzed. Prior to the survey study, we had a pilot study with both quantitative and qualitative data collected and the findings were encouraging [22]. Following this pilot-test and after doing proper modifications we could proceed to the survey study.

Data collected from the questionnaires are planned to be analyzed using path analysis (PA) [23]. The evaluated model is usually depicted in a circle (or box)-and-arrow figure in which single-headed arrows indicate causation. A regression is done for each variable in the model as a dependent on others that the model indicates are causes. The values of these weights are indicative of the effects that each variable has on another (the bigger the value, the stronger the effect). Other similar techniques to validate path models are globally referred to as Structural Equation Modeling (SEM)[24].

To date, we have just finished the data collection from the field experiment and we are in the process of analyzing the data. The multivariate analysis on these quantitative data will help us validate our research model, gain some insight to the interaction of factors affecting the adoption of the new service, and determine the extent to which it suits the needs of the library patrons as well as their intention to use it in the future. We also expect the survey findings to pinpoint any weaknesses or recommendations for the design of the prototype architecture and user-interfaces.

4. PLANNED WORK and FUTURE DIRECTIONS

After the data analysis I propose to carry this work forward by making all the necessary modifications to improve our service's usefulness and usability to its users. In addition, having found what are the most critical factors (criteria) for its acceptance and intention to use, we plan on further investigating these certain factors. There is also a need to develop more accurate, objective measuring instruments, which can be used to proceed in confirmatory research approaches. Regarding the service capabilities, I would like to explore next generation services, which can be context-aware and personalized, such as recommender systems available anywhere/anytime, as well as interaction with physical collections by means of RFID detectors.

ACKNOWLEDGEMENTS

This work is funded by the PENED'03/791 research project under the Operational Program "Competitiveness" (2000-2006).

References