[Paper Review] A Method to Support Difficult Re-finding Tasks

Gangli Liu, Ling Feng|arXiv (Cornell University)|Jan 27, 2016

Personal Information Management and User Behavior14 references3 citations

TL;DR

This paper proposes a wizard-based method to support difficult document re-finding tasks by logging user-specific attributes—such as reading time, number of pages, images, and printing frequency—during document interaction. By using a question-and-answer interface to guide users in recalling these memory cues, the system reduces cognitive load and improves re-finding accuracy, with evaluation showing effective filtering of large document sets using general memory cues.

ABSTRACT

Re-finding electronic documents from a personal computer is a frequent demand to users. In a simple re-finding task, people can use many methods to retrieve a document, such as navigating directly to the document's folder, searching with a desktop search engine, or checking the Recent Files List. However, when encountering a difficult re-finding task, people usually cannot remember the attributes used by conventional re-finding methods, such as file path, file name, keywords etc., the re-finding would fail. We propose a new method to support difficult re-finding tasks. When a user is reading a document, we collect all kinds of possible memory pieces of the user about the document, such as number of pages, number of images, number of math formulas, cumulative reading time, reading frequency, printing experiences etc. If the user wants to re-find a document later, we use these collected attributes to filter out the target document. To alleviate the user's cognitive burden, we use a question and answer wizard interface and provide recommendations to the answers for the user, the recommendations are generated by analyzing the collected attributes of each document and the user's experiences about them.

Motivation & Objective

• To address the challenge of difficult re-finding tasks where users cannot recall standard metadata like file paths or keywords.
• To reduce user cognitive burden during memory recollection by using a step-by-step question-and-answer wizard interface.
• To collect and leverage non-traditional attributes—such as reading duration, number of images, and printing behavior—stored in long-term memory for document retrieval.
• To develop a precise algorithm for estimating user reading time on PDF documents using input activity patterns.
• To evaluate the effectiveness of attribute-based filtering in narrowing down large document sets using general memory cues.

Proposed method

• Log all possible document attributes and user interaction experiences (e.g., reading time, page count, image count, printing events) during document viewing.
• Use a question-and-answer wizard interface to guide users in recalling memory cues, with recommendations generated from stored attribute data.
• Implement a reading time estimation algorithm that detects inactivity intervals to determine when reading has paused or ended, minimizing type I and II errors.
• Analyze user input interval patterns to personalize the inactivity threshold for reading time detection, adapting to individual reading habits.
• Apply attribute-based filtering using multiple general memory cues (e.g., 'over 1M in size', 'on E drive', 'printed on a weekend') to reduce candidate sets from 10,000 to ~15 documents.
• Store all data locally on the user’s device with password protection and optional deletion of sensitive records to preserve privacy.

Experimental results

Research questions

• RQ1 How can user memory cues beyond standard metadata (e.g., file name, path, keywords) be effectively used to support difficult re-finding tasks?
• RQ2 What is the optimal way to minimize cognitive load during memory recollection for document re-finding?
• RQ3 How accurately can reading time on a PDF document be estimated using input activity patterns?
• RQ4 To what extent can general, vague memory cues (e.g., 'it was over 1MB', 'I read it on a weekend') reduce the candidate document set?
• RQ5 What are the privacy implications of logging user document interaction data, and how can they be mitigated?

Key findings

• The system successfully reduces a candidate set of 10,000 documents to approximately 15 by using 16 general memory cues, each filtering out about one-third of unrelated documents.
• Over 85% of user input intervals are less than 5 minutes, indicating that most users exhibit short inactivity periods during reading, enabling accurate reading time estimation.
• User-specific inactivity thresholds (e.g., 10–20 minutes) can be derived from input pattern analysis, improving reading time detection accuracy for individual users.
• The wizard interface significantly reduces cognitive load by breaking down complex memory recall into manageable, independent questions with intelligent recommendations.
• The system demonstrates that non-traditional attributes—such as number of images, printing frequency, and cumulative reading time—are viable and effective for document re-finding.
• The evaluation confirms that even vague, general memory cues can be leveraged effectively when combined, enabling successful re-finding in difficult scenarios.