Cardinal: Novel software for studying file management behavior

Abstract

In this paper we describe the design and trial use of Cardinal, novel software that overcomes the limitations of existing research tools used in personal information management (PIM) studies focusing on file management (FM) behavior. Cardinal facilitates large-scale collection of FM behavior data along an extensive list of file system properties and additional relevant dimensions (e.g., demographic, software and hardware, etc). It enables anonymous, remote, and asynchronous participation across the 3 major operating systems, uses a simple interface, and provides value to participants by presenting a summary of their file and folder collections. In a 15-day trial implementation, Cardinal examined over 2.3 million files across 46 unsupervised participants. To test its adaptability we extended it to also collect psychological questionnaire responses and technological data from each participant. Participation sessions took an average of just over 10 minutes to complete, and participants reported positive impressions of their interactions. Following the pilot, we revised Cardinal to further decrease participation time and improve the user interface. Our tests suggest that Cardinal is a viable tool for FM research, and so we have made its source freely available to the PIM community.

The ubiquitous digital file: A review of file management research

“…Second, representative data sets (that is, test collections) do not currently exist, preventing apples-to-apples comparisons in evaluating FM systems (for example, system and user performance), and one may reasonably doubt the possibility (or usefulness) of generalized FM collections-and also of generalized PIM models-given that PIM is, by definition, such a highly personalized domain. Such efficacy may soon be empirically tested, however, as the possibility of creating such collections and models draws near: recent methodological contributions have enabled observing extensive file system properties across many participants (Dinneen, Odoni, Frissen, & Julien, 2016), thus constituting a step toward generating representative and generalized test collections, and activity logging may be used to model user behavior (Chernov et al, 2008). Third, traditional evaluation measures do not apply straightforwardly to FM contexts; for example, recall and precision are of limited use in FM retrieval evaluation, as most FM retrievals are looking for a particular file rather than a large batch of files (Fitchett & Cockburn, 2015), and it is impractical to ask a single participant to make relevancy judgments for all of their documents and invalid to ask third parties to help in this (GonQalves & Jorge, 2008b).…”

“…Nonetheless, different approaches or strategies to organizing have been identified among the varied findings, albeit rather broadly, so that we can describe organizers as: neat or messy (Boardman & Sasse, 2004), prone to saving or deleting (Berlin, Jeffries, O'Day, Paepcke, & Wharton, 1993), and prone to filing or piling (Malone, 1983), extensive filing or single folder filing (Henderson & Srinivasan, 2011), or mixing approaches (Trullemans & Signer, 2014a). To draw conclusions beyond these, studies are needed with commensurable contexts, participant characteristics, file system measures, and results reporting (Dinneen, Odoni, Frissen, & Julien, 2016).…”

“…Automated methods facilitate studying a large sample and many variables (for example, file system properties), including temporal data, but are a technical challenge to develop and implement (Dinneen, Odoni, Frissen, & Julien, 2016). Both manual and automatic collection methods require participant trust to let the researcher, possibly perceived as an expert in PIM, see their digital organization or perceived lack thereof (Barreau, 1995).…”

“…Twenty-eight properties of the file system have been examined across the studies mentioned above, regardless of the data collection method used. This includes five Dinneen, Odoni, Frissen, and Julien (2016), resulting in 40 properties available for use in FM research (presented in Table 7). Together, these properties characterize each category of FM behavior discussed above, and in smaller groups provide insight into particular actions and challenges users regularly encounter.…”

What's in people's digital file collections?

“…Various methods have been used to collect data about personal file collections, such as guided tours of users' computers (Thomson, 2015), recording participants desktops during structured tasks (Bergman et al, 2010), and taking snapshots of user's collections (e.g., Henderson, 2009). We used the last approach, employing cross-platform, opensource software called Cardinal, described and validated by Dinneen, Odoni, Frissen, and Julien (2016) and used later by .…”