Category-Based Interfaces for Mobile Search

Introduction

Mobile Internet has become an indispensable medium of information access for users around the world. It is used increasingly as a tool for communication, information gathering, transactions, and establishing one’s online presence (Taylor et al., 2008). A survey by Kaikkonen (2011) highlights the growing importance of search as a mobile information access method: the use of mobile Web search has increased significantly between 2007 and 2010, with tasks such as searching for contact information, locations and routes being even more common than on the desktop. How people look for information in their daily life is influenced by different contextual factors, such as time, location, activity and social situation (Church, Cousin & Oliver, 2012; Church & Smyth, 2009; Hinze, Chang & Nichols, 2010; Sohn, Li, Griswold & Hollan, 2008), and is affected by various technological and demographic factors (Church, Cherubini & Oliver, 2014). Moreover, mobile search has been identified as an important information access tool in various social situations, especially as a means to satisfy information needs as they arise (Church & Oliver, 2011; Heimonen, 2009). These factors should therefore inform the design of mobile search interfaces to make them fit users’ expectations and needs.

Commercial search engine providers offer search interfaces tailored for use on mobile devices. Although these interfaces can incorporate many useful features such proactive information delivery and natural language search using voice interaction, ultimately the search results themselves are often presented using ranked result lists. Studies on mobile Web search patterns suggest that users experience difficulties with these interfaces (Kamvar & Baluja, 2006; Church, Smyth, Bradley & Cotter, 2008). For example, the search data analyzed by Kamvar and Baluja (2006) shows that mobile searchers did not explore the results actively, as only about 9% of searches went beyond the first result page. Results fulfilling the users’ information need may remain unseen simply because of ambiguous queries that do not produce relevant results in the first result page. Mobile searchers also remained focused on their initial search topics; only 25% of the subsequent queries were not directly related to the first. The lack of exploration can be explained by the relatively higher cost of interactions in the mobile environment, e.g., slow loading times and overhead of browsing through result pages. The study by Church et al. (2008) highlights other problems with mobile search user experience: almost 90% of queries and nearly 60% of search sessions do not lead to any result selections by the user. It is likely that in some cases the conventional approach to mobile search adopted by search engines results in users failing to find relevant information with the result lists (Church et al., 2008).

Although current mobile search interfaces can be effective in answering focused information needs, one of the key issues with the traditional ranked result list is that it does not provide an effective overview of the result set. This makes search challenging when one’s needs go beyond simple keyword lookup, or when the query is hard to specify. Users engaging in more exploratory search of an unfamiliar topic may require additional help in understanding the terminology and structure of the result set (White, Drucker, Kules & Schraefel, 2006). Such assistance can be provided by categories, which can help information seekers make sense of search results and decide which actions to pursue (Hearst, 2006). Category-based search and browsing is commonly used in online stores where products are organized into consistent hierarchies, but it is yet to be adopted in large scale for general Web search, either on the desktop or in mobile search interfaces.