IEEE Proceedings of 4 th International Conference on Intelligent Human Computer Interaction, Kharagpur, India, December 27-29, 2012 Gardener: A file browser assistant to help users maintaining semantic folder hierarchy Debmalya Sinha Department of Computer Science and Engineering Indian Institute of Technology, Kharagpur Kharagpur, India Email: debmalya.01@gmail.com Anupam Basu Department of Computer Science and Engineering Indian Institute of Technology, Kharagpur Kharagpur, India Email: anupambas@gmail.com Abstract—Users, especially beginners, have problem finding files from large, ever growing filesystem. Common practice is to memorize the pathnames of frequently accessed folders and files which increases the cognitive load while file browsing. A suitable solution is to have a semantic hierarchy among folders so that the users can use reasoning instead of memorizing pathnames while file browsing. However, users are usually reluctant to put files in their proper places each time a file is saved. Thus, practically maintaining a Semantic Folder Hierarchy in a large filesystem is hard. This article proposes a file browser assistant “Gardener” which helps the user to maintain the Semantic Folder Hierarchy Tree in a filesystem by suggesting a suitable location each time a new file is created. The User Interface also has a “Single Click Save” option that eliminates the need to browse the filesystem and reduces interaction time. Although indirectly, this assistance for maintaining Semantic Folder Hierarchy eventually reduces the cognitive load of the users while finding files in a filesystem. Index Terms—HCI, File Browser Assistant, Semantic Hierar- chy, Filesystem Hierarchy, Intelligent Assistant I. I NTRODUCTION The number of files and folders in a computer system is ever growing[1]. Keeping track of all the files in the filesystem is an almost impossible task for the users nowadays. The highly increasing rate of the number of files results in either overflowing Top Level folders like Desktop or Downloads, or getting lost inside the deep filesystem hierarchy. As the information is increasing, the folder hierarchy tree is getting deeper, wider and more difficult to manage. While finding files is one of the most important task while using computers, the increasingly large and sparse filesystems are making it very hard for the users. A. Related Works There are a number of recent researches in Personal In- formation Management(PIM) that deals with quick location prediction while finding files. Most of them provide predic- tions based on the user specific active properties[8]; i.e: recent activities and most visited folder nodes on a filesystem tree like FolderPredictor[4] by Bao-et-al. There are also many email categorization approaches that work with related principle. Some of the works deals with the download location prediction with the browsing history, file types and URLs[5]. The major limitation with these approaches is the users mostly browse through the hierarchy while finding files and they seldom use any finding tools for that[7]. Moreover, the heavy reliance on user behavior and past history of searching requires a significant time to be spend with computers which a novice user do not have; making these approaches restricted only for experienced users. B. Our approach This paper approaches the problem of finding files from a different perspective rather than concentrating on folder loca- tion prediction based on user behavior. The earlier studies[6], [7] about user behaviors show that, albeit the existence of automated file finding programs, users prefer to browse the filesystem by contextual local steps while finding files. Based on this, we’ve proposed to build up a semantic hierarchy in the filesystem to ease the process of finding files. In such a hierarchy, any user will be able to browse by reasoning through various hierarchical contexts in the filesystem. Users essentially have two main techniques to find files. They either use their Spatial Memory to remember the lo- cations of files; or they use semantic reasoning between hierarchical concepts in a folder hierarchy to find the lo- cation of a specific file. A complete reliance on memory is impractical while working with a large filesystem. It is very common to forget about not-recently-used existing files due to memory decay. Instead, using semantic reasoning is much more convenient, given the low cognitive load for logic processing of human brain. Users can browse by relevant hierarchical contexts using simple reasoning regardless the size of the filesystem. However, to enable semantic reasoning, the filesystem must be a Semantic Folder Hierarchy itself. The first problem implementing such a Semantic Folder Hierarchy is its maintainability. Users have to maintain it each time there is a new file or folder added by putting them into their befitting contextual positions in the filesystem. In practice, the users are usually very reluctant for this and they simply store every new file in some top level folders like Desktop or Downloads[6] which deranges the semantic folder hierarchy. To maintain it, users need an intelligent assistant which will help them maintaining the semantic hierarchy by automatically suggesting a befitting position every time a new file is saved. As a well maintained Semantic Folder Hierarchy is the key to contextual reasoning based browsing; eventually, 978-1-4673-4369-5/12/$31.00 c 2012 IEEE