Geographic information retrieval in a mobile environment: evaluating the needs of mobile individuals

Abstract

This is the accepted version of the paper.This version of the publication may differ from the final published version. Permanent repository link AbstractIn this paper, the information needs of mobile individuals are contrasted with those of their static (desktop) counterparts. The information needs of users of a mobile information system, implemented in an outdoor recreational area, are first investigated in a two-part user needs study, which finds a strong geographic component to their information needs have a strong geographic component. Next, four geographic post-query filters which are described which attempt to meet these needs. These filters are spatial proximity (distance in space), temporal proximity (travel time), speed-heading prediction surfaces (likelihood of visiting locations) and visibility (locations that can be seen). Two of these filters (spatial proximity and speed-heading prediction surfaces) are implemented in a mobile information system are evaluated with users in an outdoor setting. The results of evaluation suggest that information that to which post-query geographical filters have been applied is more relevant than unfiltered information, and that users find information sorted by spatial proximity to be more relevant than that sorted by a prediction surface of likely future locations. The paper closes by suggesting that one of the contributory factors to the failure of location-based services to become widely adopted, could be a neglect of the distinct information needs of mobile individuals.

Geographic dimensions of relevance

“…• activity: all the activities related to the information need, ranging from those that generated the information need to those that will be pursued on the basis of the retrieved information (this component deserves some further explanations, see below); • preferences: user's preferences related to the topic or the activity considered (e.g., Rashid et al, 2002); • social: the social context that can influence the relevance of an entity, such as the popularity within a community or among user's connections (e.g., Mizzaro and Vassena, 2011); • mobility: the spatio-temporal availability of the entity (e.g., location and opening hours), the spatiotemporal situation of the user (e.g., location, time schedule, and mode of transportation), available mobility infrastructure, and topological structure (e.g., Mountain and MacFarlane, 2007); • context: everything not pertaining to the previous components, including knowledge about the physical surroundings (e.g., light level or other geographic entities in the surroundings) and informational surroundings (e.g., information contextual to the topic or activity).…”

“…Turning to the following three rows in the table, we can note that MIR systems (see second row in the table) usually include the mobility component (Mountain and Macfarlane, 2007) and generally account for the spatio-temporal point at which the query is submitted to the system (see the 'space-time' column), e.g., providing references to relevant documents to a user exploring a national park, Mountain and Macfarlane, 2007. Relevance as it is commonly implemented in LBSs (see third row in the table) can be defined as being concerned with features and the user's query, again at the spatio-temporal point at which it is submitted to the system (e.g., providing points of interest to a user hiking in the mountains, Hauthal and Burghardt, 2012).…”

“…As mobile scenarios become more and more common, it is important to analyse how different relevance concepts deal with a user's mobility (Mountain, 2005;Mountain and MacFarlane, 2007;Raper, 2007;Raper et al, 2007), and represent the world surrounding a user (Reichenbacher, 2004;Reichenbacher and De Sabbata, 2011;De Sabbata and Reichenbacher, 2012). Relevance has always been recognized as a crucial notion for IR, and there is a vast body of literature devoted to it.…”

Gander: Personalizing Search of the Here and Now

“…Recent work has assessed requirements for information retrieval and identified relationships between users' information needs and their context, including location, time, and even activity [17,23]. Such studies have exposed influences on people's willingness to share information about themselves; for example, people may be willing to share with other co-located users (even unknown ones) what they would not be willing to share publicly (i.e., on the Internet).…”

“…WebPark [23] provides geographically relevant information using geographic filters to capture spatial relevance. Ahlers and Boll enrich spatial context with temporal aspects [1].…”