User Evaluation of a Prototype for Tagging and Sharing Mobile Contacts Trung, Nguyen Van

Alice H. Oh

KAIST Computer Science Daejeon, Republic of Korea

KAIST Computer Science Daejeon, Republic of Korea

[email protected]

[email protected]

ABSTRACT We present the results of a user evaluation of a prototype for tagging and sharing mobile contacts which we proposed as a solution to the problems with existing contacts application informed by our previous study [17]. Such problems include the inflexibility of organizing with groups, the inability of users to remember contact names for retrieval, especially in the case of not frequently interacting contacts, and the difficulty in determining the right people to ask regarding information about business contacts. This paper describes the user study that we conducted as some first preliminary design considerations that we planned to build into the system. Overall, users were positive about the system, and their useful feedbacks were already incorporated into development of the application that we almost finish developing for testing the applicability of tagging and sharing contacts.

Categories and Subject Descriptors H.5.2 [User Interfaces]: User-Centered Design

General Terms Experimentation, Design, Human Factors

Keywords Mobile contacts application design, user study, paper prototype, mobile tagging, recommender system

1. INTRODUCTION The mobile contacts application has been a relatively inactive research problem in the past. However, recent years have seen an increase in academic research on designing the future mobile contacts application, perhaps due to the new generations of smart phones with many features such as WiFi. In this paper, we summarize our previous research in which we discovered users’ dissatisfaction with managing contacts and looking for information beyond their own contacts such as business services. We propose tagging contacts and sharing the tags with one’s social network as a solution to these problems. We present user evaluation of the proposed system in a paper prototype study which showed that tagging and social sharing combined can potentially eliminate the problems and positively enhance user experience with the system. We also discuss future research directions in potential applications of combining social tagging and sharing on the mobile phone.

1.1 Mobile contacts application Traditional features of the mobile phonebook mainly include adding a contact, searching for a contact, and organizing the Copyright is held by the author/owner(s). MobileHCI 2010 September 7 - 10, 2010, Lisboa, Portugal. ACM 978-1-60558-835-3. .

contacts list. Surprisingly, little research has studied the usability for these user activities on the phonebook despite it being one of the most used applications on mobile devices. In one study, Komninos and Liarokapis [10] attempted to facilitate access to important contacts indicated by high frequency and recency of use. In another study, Jung et al [8] implemented a prototype that included efficiency of access, differentiation of important contacts, customization and personalization, flexibility in organization as the design drivers. The main changes made were embedding groups in the list and making them searchable, differentiating top 10 contacts manually or automatically, and automatically forming Smart groups such as Popular (top communicators), Birthday, and Shiny new (contacts added recently). Both research projects lacked a usability study on main activities (storing, searching and organizing contacts); hence problems that might occur when users conduct these activities remained untouched. We found out in our earlier user study with 87 users that problems did exist with the main adding, searching, and organizing mobile contacts [17]. We believe that providing solutions for them is important because these are the tasks that are most often carried out by users on their contacts application. Some previous research focused on providing context awareness for the mobile contacts application. In [15], Schmidt et al implemented a phonebook that made contextual information including device connection status (switched off, on the phone), available preferences (willingness to receive call in current situation), and current location available to users. In a similar study, Oulasvirta et al [14] presented the design of a mobile awareness application which provides cues of current situation including location, time spent at that location, user-selected alarm profile, recent phone manipulation, and several other information. Although both studies are not directly related to our study, we realize the importance role that context may play in our future system such as improving tag suggestions. More recently, Bentley, et al [2] introduced the Contacts 3.0, an application that serves as a single destination for multiple communications and provides a sense of shared experience. The shared experience included was photo and music, but our study result found out that sharing of business services information is also important to and useful for users.

1.2 Tagging Content tagging is a very active research problem in recent years. Many systems exist that allow a variety of contents to be tagged such as photos by Flickr, webpages by del.icio.us, videos by Youtube, and blogs by Technorati. These community tagging systems are widely available as web services. Several systems exist within the PC-based environment including TAGtivity and

Phlat [13]. However, tagging on mobile platforms has not yet become as popular as the other two platforms, and to our knowledge, ours is one of the first looking at tagging on a native mobile application for tagging contacts. Using tags in their mobile phonebook, users can annotate people, services, and businesses; and some those tagged contacts, such as the public and business services, can be made available for community sharing. There is also research on people’s motivations for tagging based on qualitative and quantitative study [1, 12] as well as cognitive psychology [7]. One argument is that tagging can be effective for organization and communication of personal and sharable resources - search and retrieval, memory and context, public search, context and signaling [12]. Another argument is that tagging can be a valuable addition to traditional hierarchical organization method due to its better fit with the working of the human memory than rigid folders [7]. These findings serve as the foundation upon which we believe that tagging contacts will offer similar but unique benefits to users due to the distinctiveness of content.

1.3 Personalized recommender systems A thorough review of personalized recommender systems is beyond the scope of this paper, and a good survey can be found at [3, 6]. Basically, personalized recommenders system suggest resources that can be categorized as content-based (previous resources consumed by users), familiarity-based (resources consumed by people that users know), and similarity-based (resources consumed by people that are similar to users) [6]. The study by Guy et al [6] suggests that familiarity-based recommendation outperforms the other two approaches, but it applied to social software items only. The information that users provide to our proposed mobile contacts application is clearly small compared to other systems; hence it will be interesting to evaluate the performance of familiarity-based recommendation for business services on the system.

2. TAGS AND SOCIAL NETWORKS In this section, we briefly summarize the findings of our previous research where we identified many problems users encounter in organizing and searching for contact information using existing mobile phone book applications. The user study and problems identified are described in more detail in [17]. We then explain our motivations for adding the social tagging and sharing features in our proposed mobile contacts application.

2.1 The needs for social tagging and sharing Our survey results showed that users have various problems with organizing their mobile phone book [17]. They often don’t know which group to put a contact into, they sometimes want to place a contact into several different groups, they cannot come up with a representative name for the group, and they complain that they cannot search using group names. We concluded that users’ need to associate, store and search for various pieces of information with a contact was not well accommodated by existing mobile phone books. We also identified users’ problems when looking for the phone numbers of business services which are important to users due to their long-lasting and repeated uses. Looking for the numbers in the mobile phone book has the problems of not remembering whether the phone number is stored, and not remembering the exact name under which the phone number is stored. Asking acquaintances, which is the second most often used method to get information about new services, has the problems that the users often do not know who to

call for the information, and when they decide to call someone, that person often does not have the information.

2.2 Storing, organizing and searching tags Our survey revealed that users attempt various solutions on their own. 7 participants used memo or note for storing the extra information, and several others include the information in the name entry of a contact. Both suggest the need for easy and efficient storage and retrieval of extra information related to a contact. Yet, both memo and note are not available for searching in the contacts application as the contact name has been the only searchable piece of information for a contact. The solution attempted by the users is indeed related to what we believe to be an inherent solution to this problem: adding tags for each contact. The act of tagging a resource is similar to bookmarking a resource for oneself, and tags are useful for retrieval of resources [5]. Resources here are contacts. Thus, associating tags with a contact proves to be a good way for users to easily search for it later without having to remember name of the contact or browsing the entire contacts list. Moreover, Millen et al [11] suggested that resources can also be categorized by several tags, rather than one directory or a single branch of a hierarchy. The act of assigning a contact into multiple groups then becomes the simple task of adding multiple tags to that contact. Hence, tagging might help satisfy users’ basic need for efficiency of access, flexibility in organization, and customization and personalization with minimum efforts.

2.3 Social network based search for business services Even though tagging can improve the storing, organizing and searching experience, it does not solve the problem that users have when they need information for a new service. Our survey results show that users often turn to their acquaintances for information, and they need help in finding the right person who has the information needed. We propose combining tagging and sharing of information with the people in one’s mobile contacts. Many tags submitted by multiple users taken together can result in a complex network of inter-related users, resources, and tags [16] which can be used to infer the people with knowledge about business services that they tagged as well as generate search results for these services.

3. PROTOTYPE EVALUATION 3.1 User study with paper prototype We designed a paper prototype of the system for a user study with the following objectives: 1.

2. 3.

Understand users’ perceptions of and interactions with the system with a particular focus on the proposed new features. Evaluate the user interface of the system. Validate users’ problems identified in our previous study and users’ motivation for using our new system.

At the beginning of each study, the author explained to the user new features introduced to the contacts application. Users are then asked to perform four tasks by interacting with the paper UIs. One researcher played the role of computer by responding to users’ interaction with corresponding UIs on paper. Example paper UI

screens can be seen in Figure 1, 21. After each task, users evaluated the system by filling out the task evaluation form. We also communicated with users during and at the end of each task

to find out more about users’ problems and their motivations (objective 3).

Task 1 and task 2 focused on evaluation of the system with the capability for tagging a contact and searching by tags. In task 1, users had to follow 5 scenarios, and each scenario required them to store the contact information of a person or a service they just encountered. In addition to the contextual information of the encounter including place, reason and time, information about each contact such as name, phone number, address, job title or position, company was provided in each situation. In task 2, users were given 5 scenarios in which they had to contact the people that they had stored in task 1, though in a different order of appearance. Task 3 and task 4 focused on social tagging and sharing for business service contacts. The researcher explained to users the system’s new capability of sharing and searching for business service contacts with tags. In task 3, two scenarios with business services from task 1 were presented again to participants. They were then asked to specify the tags that they would use for themselves, tags that they would want to share with friends, and tags that they would want to share with everyone. In task 4, users were given 3 new scenarios each described a situation when they need to look for new business services with the system.

Figure 1: “Add Tags” screenshot with tag suggestions Table 1: Summary of tasks Task

Task description

# of scenarios

1

Storing new contact information

5

2

Looking up existing contacts (which were stored in the previous task)

5

3

Sharing tags (with friends and with anyone) for business services

2

4

Searching for new business services

3

Table 2: Example scenarios

Figure 2: “Contact info” screenshot for a business service Two Korean graduate students, one male and one female, one Vietnamese female undergraduate, and one Chinese male undergraduate participated in our lab-setting study. We asked them to do the study on different days so that feedback from previous users would be considered and incorporated into the system if appropriate.

3.2 Task descriptions Participants were asked to perform 4 tasks, and each task contained several different scenarios. We paid special attention to make the scenarios as close to users’ real-life experiences as possible by using techniques such as including real people names, places and relevant contextual information. A summary of the tasks and example scenarios can be seen in Table 1, 2. 1

Note that the images here are electronic version of the UIs for better image quality. In the study, hand-drawn UIs were used instead. All paper UI elements are preserved.

Task 1, scenario 3: “Last Saturday, you were preparing for your midterm exam, so you had to stay up late until 2am in the morning. You felt hungry and decided to order chicken + pizza for a night meal with your roommate. Both of you didn’t know any food delivery service, so you asked your friend Jiwon. She gave you the number 532961 of the Dota company whose business hour is until 3am. You called and ordered a chicken + pizza set priced 15000won (Korean currency). The food was tasty and you thought the price was quite acceptable. You think you would use the service again, so you decide to store the service contact in your phone book”. Task 2, scenario 5: “It’s two weeks after the midterm exam, and you are having a big project, so you have to stay up late until 2am again. You feel hungry, and you want to call the chicken and pizza service again”. Task 3, scenario 2: same as task 1, scenario 3 except the task requirement. Task 4, scenario 1: You’re holding your birthday party this afternoon. You went to the market and bought the necessary items yesterday. While preparing, you realize that you forgot to buy fresh strawberries. You estimate that you need 6 boxes of strawberry for 15 people. You want to call a service and order the strawberries for delivery..

3.4 Improvements from user feedback 3.3 Usage and evaluation of system For the evaluation of the system to be valid, the users’ interaction with the system should resemble their real world experiences as much as possible. All users commented that the scenarios were realistic and they had encountered similar situations many times in their lives. They also thought that the scenarios had good coverage of the information they generally need when dealing with contact information.

In addition to their positive evaluation, users’ suggestions and feedback also allow us to make some improvements to the system. One such improvement comes from the result of task 1 which indicated that name, phone number and tags are the three most frequently stored information. Email, company name, department and job position were less frequently stored but considered important for professional contacts. This led us to a simplified design for the “Add new contact” screen shown in Figure 3.

Overall, users showed positive feelings towards the system. For example, one user was happy with the feature of tagging in the system and actively used it when she stored the new contact information because she experienced quite many times the problems of unable to recall names of people when in need which she normally gave up. Another user commented that the system would be helpful especially when he goes out. Additionally, the researcher observed users’ facial expression and communicated with users when they interacted with the system; only a few small confusions with the UI elements, which will be discussed in more details in the next section, were noticed. Most importantly, we observed all of our intended purposes for the system in this small set of users’ system usages. First, users often associated with the contacts the key information, personal experiences or impressions either by using tags or enclosing the information within the name field. In all instances, they explained that the information would make search easier in the future. In fact, users only used the name of the person or service as search keyword once out of the 20 scenarios given in the second task; yet they regularly queried using information from the scenarios that was memorable and personal. This suggested that the system helped solve users’ problem by enabling them to store retrieval cues - tags, for easy recall of contacts. Second, users were happy with the help from the system in finding new business services. The basic information for a service (Figure 2) was enough for some users to make decision on whether to call a service or not. One user, however, relied on the system to get the list of people in his network that tagged the services and would call them to inquire more information. This showed that the system’s capability in recommending the right person or service was useful for and readily received by users.

Figure 3: Simplified “New contact” UI screen The most critical feedback for UI improvement to the system was, however, related to categorizing and organizing contacts with tags. One user wanted to categorize the newly added contacts but found it difficult to accomplish using the given UIs. We realized that the UI elements were not intuitive enough for users to figure out that tags can be used for organize the contacts. Figure 4 shows the screenshot of our implemented system that addresses this issue by displaying each contact with its associated tags which can be colorized and selected.

We also discovered some limitations with the system. Users conceded that they are often lazy to input text on their mobiles although they know that certain information about a contact is really important for them. We all agreed that this often comes with the cost of unable to find a contact when in need. Sometimes the cost was huge because users had to try to recall every person in his contacts list to find the contact, yet still failed to recall. Gradually, this cost can shape users’ habit of trading off time to input for convenience and efficiency in the future. In addition to this incentive of tagging a contact, a good tag suggestion system can save users’ effort in inputting tags. We discuss the tag suggestion system in more detail in section 4.1. Another limitation is that the goal of using tags as a flexible and efficient way for organizing users’ contacts was not easily noticeable to users. This could be due to the limited amount of time and contacts that users interact with the application during one session. However we still consider this one point for improvement as discussed in the next section. Figure 4: Contacts with associated tags

4. IMPLEMENTATION ISSUES Going from our paper prototype to actual implementation should be straightforward for the most part, but we identified two important issues for the system to be widely adopted by users: tag suggestions and business service recommendations.

4.1 Tag suggestions Inputting text on mobile devices is a time consuming and cumbersome task [9], and users are not always willing to put in the time and effort. The user study revealed that there is enough motivation for the users to use tags for solving their problems, but providing good tag suggestions will make the system friendlier and easier to use. In a related task of mobile search, Kamvar et al [9] suggested that query suggestion was shown to lower users’ workload and increase their enjoyment. Additionally, the ZoneTag application introduced in [1] demonstrated that offering tag suggestions increased the number of tags. This is also supported by our users’ positive feedback about having tag suggestions during their task of adding new contacts. In the system that we are developing, candidate tags for suggestion are generated from aggregation and ranking of different sources including tags that are often used together and user’s frequently used tags.

4.2 Social search for shared contacts In task 4, we showed each user different combinations of user ratings (high or low) for services with relationships (close or remote) between the user and the people in their social network who gave the ratings. We then asked users for their preferences in choosing the recommended services. Their responses suggested that, in general, a user rating is a good indicator of the quality of a service regardless of the source of the rating. When there is no significant difference in ratings, users would either call people or make decision based on “trust”, which can be his knowledge about the other party’s familiarity with a particular type of service and/or his relationship with that person. Considering that ratings are not always available, a person’s social network becomes a critical source of recommendation. Indeed, social network based personalization search was shown to outperform non-personalized social search in a study by Carmel et al [3], so we plan to apply and evaluate a similar approach for our business service recommendation system.

5. CONCLUSION AND FUTURE WORK We discussed problems in the current mobile phone book application and users’ evaluation of a paper prototype with social tagging and sharing as a solution. The user study results showed that users were positive with the proposed features and the system. Our short-term goal is to build and evaluate a real system based on the paper prototype with the implementation considerations discussed in section 4. We have identified several interesting issues and research goals for long-term future work. The first goal is to reduce users’ workload when adding and tagging a new contact. One possible research could be to capture the users’ context and physical surrounding to improve the accuracy of tag suggestions. Another research issue is on privacy regarding the sharing of personal contacts. Users are generally concerned with the idea of sharing their contacts, but people-tagging and sharing in the enterprise was well accepted according to [4]. We thus are interested in studying more about the users’ willingness to share their contacts: under which conditions, with whom, and what aspects of the

contacts they would be willing to share. We also want to examine the potential of integrating our system with that of the Contacts 3.0 or similar applications. Social tagging and sharing with user’s multiple social network services (SNS) such as Facebook, Twitter, Flickr can create some useful experiences for users. These SNS normally only allow users to share with general network or specified groups of people such as friends, or friends of friends, or public. Using tags can make sharing and communication more selective and dynamic, for example, by allowing users to share or communicate with a specific group of people tagged with the same word.

6. REFERENCES [1] Ames, M., Naaman, M. Why we tag: motivations for annotation in mobile and online media, CHI’07, pages 971980, 2007. [2] Bentley, F., Kames, J., Ahmed, R., Zivin, R.S., Schwendimann, L. Contacts 3.0: Bringing together research and design teams to reinvent the phonebook, CHI2010, 2010. [3] Carmel, D., Zwerdling, N., Guy, I., Ofek-Koifman, S., Har’el, Nadav., Ronen, I., Uziel, E., Yogev, S., Chernov, S. Personalized social search based on the user’s social network, In Proceeding of the 18th ACM conference on Information and knowledge management, pages 1227-1236, 2009. [4] Farrell, S., Lau, T., Fringe Contacts: people-tagging for the enterprise, In Proceedings of the Collaborative Web Tagging Workshop in conjuction with the 15th International Conference on the World Wide Web, 2006. [5] Furnas, G.W., Fake, C., von Ahn, L., Schachter, J., Golder, S., Fox, K., Davis, M., Marlow, C., and Naaman, M. Why do tagging systems work? In CHI Extended Abstracts, pages 3639, 2006. [6] Guy I., Zwerdling, N., Carmel, D., Ronen I., Uziel, E., Yogev, S., Ofek-Koifman, S. Personalized recommendation of social software items based on social relations, RecSys’09, pages 53-60, 2009. [7] Hsieh, J.L., Chen, C.H., Lin, I.W., Sun, C.T., A web-based tagging tool or organizing person documents on PCs, In CHI 2008 Workshop on Personal Information Management, 2008. [8] Jung, Y., Anttila, A., Blom, J., Designing for the evolution of mobile contacts application, MobileHCI’08, pages 449 – 452, Amsterdam, The Netherlands. [9] Kamvar, M., Baluja, S. Query suggestions for mobile search: understanding usage patterns, CHI’08, pages 1013-1016, 2008. [10] Komninos, A., Liarokapis, D. The use of mobile contact list applications and a context-oriented framework to support their design, MobileHCI’09. [11] Millen, D., Feinberg, J., and Dogear, B.Kerr, Social bookmarking in the enterprise. CHI’06, pages 111-120, 2006. [12] Nov, O., Naaman, M., Ye, C. What drives content tagging: the case of photos on Flickr, CHI’08, pages 1097-1100, Florence, Italy, 2008. [13] Oleksik, G., Wilson, M.L., Tashman, C., Rodrigues, E.M., Kazai, G., Smyth, G., Milic-Frayling, N., Jones, R. Lightweight tagging expands information and activity

management practices, CHI’09, pages 279-288, Boston, USA, 2009. [14] Oulasvirta, A., Raento, M., Tiitta, S. ContextContacts: Redesigning smartphone’s contact book to support mobile awareness and collaboration, MobileHCI’05, pages 167-174, 2005. [15] Schmidt, A., Stuhr, T., Gellersen, H. Context-Phonebook – extending mobile phone applications with context, Third Mobile HCI Workshop, 2001

[16] Shepitsen, A., Gemmell, J., Mobasher, B., and Burke, R., Personalized recommendation in social tagging systems using hierarchical clustering. In Proceedings of the 2nd ACM Conference on Recommender Systems, pages 259 – 266, 2008. [17] Trung, N.V. and Alice, Oh. Users’ needs for social tagging and sharing on mobile contacts. Submitted to MobileHCI 2010.