Automatic Assignment of Topical Icons to Documents for Faster File Navigation

Several computer users neither assign names to their documents systematically nor organize them into suitable folders, making it difficult to search for relevant files when needed. While this problem can be addressed in several ways, we explore the novel approach of automated assignment of topical i...

Full description

Saved in:
Bibliographic Details
Published in2017 14th IAPR International Conference on Document Analysis and Recognition (ICDAR) Vol. 1; pp. 1338 - 1345
Main Authors Roy, Rishiraj Saha, Singh, Abhijeet, Chawla, Prashant, Saxena, Shubham, Sinha, Atanu R.
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.11.2017
Subjects
Clustering algorithms
Document iconification
Electronic mail
File search
Finance
Navigation
Software
Task analysis
Taxonomy
User studies
Online AccessGet full text

Cover

More Information
Summary:Several computer users neither assign names to their documents systematically nor organize them into suitable folders, making it difficult to search for relevant files when needed. While this problem can be addressed in several ways, we explore the novel approach of automated assignment of topical icons to documents in order to cue memory for faster navigation. Specifically, we overlay the currently available generic software-oriented file association icons like Acrobat, Word or Powerpoint on documents with algorithmically assigned icons that are specific to the topical content of the documents. Our pipeline method uses document clustering, significant-phrase extraction, phrase generalization and phrase vector matching for assigning icons to documents. Experimental results show that topical iconification significantly speeds up document navigation time vis-á-vis content-based file naming, in both a controlled laboratory setup as well as in a crowdsourced study. Icons assigned by our algorithm are observed to have satisfactory inter-annotator agreement with respect to their meanings.
ISSN:2379-2140
DOI:10.1109/ICDAR.2017.220