... puts the fun back in file navigation

This page contains tagstore-related scientific papers and thesis. You can find raw data sets and derived data on the web:

TagTrees: Improving Personal Information Management Using Associative Navigation

This is a PhD thesis from Karl Voit which summarizes most of the research work done from 2009 to 2012 within the tagstore project. Its chapter four describes the TagTrees navigational structure in detail. The chapter five can be seen as a manual for the tagstore software.

Abstract

This dissertation gives an overview of research related to Personal Information Management (PIM). After discussing some challenges that are important to be addressed, a new method for managing local files is described: TagTrees. Navigational structures for re-finding are automatically generated in the file system after associating tags to items (files or folders). These TagTrees allow for associative navigation, because each item is reachable via a large set of navigational paths in the file system hierarchy. The TagTrees method was implemented in a research software called tagstore. This implementation provides multiple configuration possibilities, which qualifies tagstore to be a research framework for analyzing many different kinds of tagging processes. Besides long-term usage patterns, two formal experiments were conducted to evaluate tagstore. Comparative studies showed that tagstore performed similarly to the traditional folder hierarchy method. Subjective user feedback and acceptance was very positive. Having a comparable performance for filing, the additional possibilities for re-finding items combined with very high user acceptance makes TagTrees and tagstore a valuable contribution to PIM research and end user software. In order to reproduce research results and re-use the software, the tagstore framework itself, experiment artifacts of all kinds, study results and evaluation tools are made public under an open license.

Downloads: PDF; BibTeX

Tagging Might Not Be Slower Than Filing in Folders

The CHI is a well established conference for the Human Computing community, having a high rating. Preliminary results of one of our formal experiments are the content of this paper. The content of this paper was presented as a poster on the CHI2012 conference in Austin, Texas.

Tagging is a promising method for organising and re-finding information. However, studies comparing tagging mechanisms to organising information in folder hierarchies are relatively scarce. A study with the software framework tagstore shows that tagging does not necessarily mean slower filing performance. For experienced users, tagging required less time, fewer mouse clicks, and had very high acceptance rates.

Downloads: PDF; BibTeX

Formal Experiment Report: Tagging Files vs. Placing Files in a Hierarchy

This is a report with the summary and the most important results from our first formal experiment where we compared the subjective and objective performance of tagstore and the usual method of filing and re-finding with Windows Explorer.

In general, filing in folders seems to be faster than filing in tagstore. For a sub-group of so called "fast performer", filing in tagstore was faster than filing in a folder hierarchy. For re-finding, there is no significant difference between the two conditions. Subjective feedback was overwhelming positive. For example 78 percent of the participants said that they would prefer tagstore over their current method using a strict hierarchy of folders.

It seems to be the case that tagstore offers much more possibilities to manage items (files and folders) than the traditional method with no significant additional cost and a huge benefit in user acceptance and fun.

Downloads: PDF from github repository; BibTeX; github repository

A summary of this report with additional statistical analysis is part of this PhD thesis: "TagTrees: Improving Personal Information Management using Associative Navigation" (Karl Voit, to appear in 2012-11).

Creating a Comparative Environment for PIM Evaluation

PIM Workshop is the most important event for the PIM community. After attending 2009 I was very happy to present another paper in 2012. The focus in this paper is the importance of having really comparable software products for testing.

A common form of comparative evaluation in current research is to run a formal experiment with a number of test users and software implementations of two or more underlying research approaches. When the underlying methods are, in fact, the subject of the comparison, it is essential for fair comparison that their software implementations and interface usability be of equivalent quality. Previous work on evaluating information visualisation techniques can inform such evaluations in the area of human-computer information retrieval. A research framework called tagstore supports experiments on different styles of tagging interfaces. It can also be used for experiments involving the storing and re-finding of personal files using either folder hierarchies or tagging. The tagstore framework provides a basis for comparable results, flexible control of appearance and functionality, and multi-platform availability.

Downloads: PDF from PIM2012 webpage; Author version of PDF; BibTeX

Cited by others: Google Scholar; Google Search

TagTree: storing and Re-Finding Files Using Tags

At USAB 2011 I had the pleasure to present this paper which describes the basic principle of the Method (TagTrees) as well as a short description of the implementation (tagstore).

Although desktop search engines are now widely available on the computers of typical users, navigation through folder hierarchies is still the dominant mode of information access. Most users still prefer to store and search for their information within a strict hierarchy of folders.

This paper describes TagTree, a new concept for storing and retrieving files and folders using tagging and automatically maintained navigational hierarchies. TagTree is compatible with all currently prevalent software environments. A prototype implementation called tagstore provides a flexible framework for experimentation and a testbed for both usability studies and longer term field tests.

Preliminary test results show a very positive user acceptance rate of using TagTrees for storing and re-finding files.

Downloads: PDF at Springer; Author version of PDF; BibTeX

Cited by others: Google Scholar; Google Search

Why Personal Information Management (PIM) Technologies Are Not Widespread

Accepted at PIM09 this paper summarizes a research why advanced Personal Information Management Tools are not on everybodys desktop. In this paper, tagstore existed only as an early prototype for technological scouting.

Users of computer systems create and store valuable personal information in files, email folders, and bookmark collections. For decades, the main principle of interacting with files, emails, and bookmarks has remained unchanged: hierarchical directory trees with standard (Windows Explorer style) browsers.

Users often have problems both in classifying new items and maintaining a classification hierarchy as such. With files, emails, and bookmarks, users often end up maintaining three parallel classification hierarchies, one in each tool. Over the past thirty years, a number of alternative personal information management (PIM) tools have emerged, but the typical user is still faced with hierarchical directory structures.

This position paper addresses some of the reasons why modern PIM tools are not widespread and proposes a set of eight requirements for future PIM tools.

Downloads: PDF; BibTeX

Cited by others: Google Scholar; Google Search

Marktübersicht von Tagging-Werkzeugen und Vergleich mit tagstore

This bachelor thesis contains a very detailed market analysis for software tools that are using tagging for managing files.

Based on the increasing amount of information that every computer user has to manage, efficient information management is necessary. By using efficient information management, it’s possible to reduce a high percentage of wasted time. Tagging is an effective instrument for classifying and re-finding information. At Graz University of Technology, a research platform called tagstore has been developed. This platform makes it possible to link files and folders with tags. This new form of information management has many advantages. In this thesis, numerous tagging tools were tested, evaluated with standardized methods, and finally compared with tagstore. To give an impression of the evaluated tools to the reader many screenshots were added. At the end of this thesis a table compares essential functions of tagstore“ to the other tools.

This document is in German language only.

Downloads: PDF; BibTeX

Automatic Evaluation of User Experiments using Intermediate Language and Compiler

For our lab experiments (formal experiment 1 and 2) we were using an intermediate language for the video transcripts and the test user actions. Using the compiler described in this bachelor thesis, we were able to process the data automatically. Input files are the (manually generated) transcription files. Output files are CSV files containing raw data and some derived data (mean values, ...). Such a method is crucial for Reproducible Research and Open Research.

Since usability evaluation has become an important element of software development, it is crucial to manage and process all generated usability testing data efficiently. This thesis compares two approaches in the field of evaluating HCI user experiments. It is shown, which input data, generated during these user experiments, is needed and which results can be generated in order to automatically evaluate qualitative and quantitative aspects. In this work the functionality of the developed tool-kit is shown with the aid of the conducted formal experiments for the software tagstore. Thereby the process of automatic evaluation of a user logfile is described in detail. It is pointed out, that a general user logfile and a finite state machine for every use case are needed to be able to evaluate this logfile automatically and time analysis and other desired results can be derived. This thesis also describes considerations regarding the re-usability of this tool-kit and automation of the evaluation process.

Downloads: PDF; BibTeX

back to home

Last modified: 2016-02-26T19.22.44