Research support with optical character recognition apps
Download as PPTX, PDF2 likes820 views
The document discusses the development of the Text-Shot prototype, which uses optical character recognition (OCR) to recommend library resources to students. It reviews various OCR apps, outlines the methodology of user testing with first-year university students, and presents results indicating desired features for enhanced recommendation capabilities. Future directions include integrating OCR technology into additional library applications and exploring open-source options for library settings.
Research support with optical character recognition apps
- 1. Research support with optical character recognition apps Jim Hahn
- 3. Introduction • Uses for OCR in library settings – The prototype Text-shot module uses OCR software and a backend search system for subject and title recommendations. – The choice to recommend library content to users from the app stems from the objective to connect students with library resources, and to help students integrate library resources into their work. 3
- 4. Optical Character Recognition Apps • Wordlens app: can translate words from different languages using a digital camera feed • Google Goggles app: take a picture of a book cover (or painting)to run a google search on the topic • Camscanner app: digitize print documents with camera on app and store/share documents with others 4
- 5. Literature Review • Optical Character Recognition APIs – Evernote API: dev.evernote.com/doc – Google Drive API: support.google.com/drive – VuForia SDK: developer.vuforia.com/resources/sdk/android 5
- 6. Methodology • Formative evaluation – Small set of test participants to gather feedback early in the design phase so that the software development process can progress in a direction that will support user requirements for the software 6
- 7. Methodology • Test Participants – Students were recruited from the General Studies 101 course. They are in their first year of study at the university and have not yet chosen a major. – There were a total of five test participants in the first round of study. 7
- 8. Methodology • Study Process – Students were given an Android phone with the Text-shot app loaded. Investigators observed the students as they used the OCR mobile software to obtain suggested library resources. Investigators collected two sources of data: observation of how students interact with the software and a debriefing interview. 8
- 9. Functionality Tests • Researchers tested the two main functions for the software. – Recognizing a string of text by taking a picture of the words in a student assignment sheet and; – suggesting subjects and titles based on the scanned text. 9
- 10. Results • Themes related to the improvement of suggestions: – Show broad subjects first • Then expand to details subjects – Prominently display title suggestions 10
- 11. Results • Feature Requests: – Include articles as well as book titles in recommendations • Use article APIs • LibGuides-like help guides 11
- 13. Next steps in OCR • Topic Space app: Scanning call numbers in the library – If you scan a call number on a book, you can get recommendations of other, related books in the library, and other related digital content in the library. 13
- 14. Topic Space: Book Scan 14
- 15. Topic Space: Suggested Topic Spaces 15
- 16. Topic Space: Related Books that are not available 16
- 17. Topic Space: View Map 17
- 18. Future directions • Implementing OCR modules in the Minrva app: – http://minrvaproject.org/modules_topicspace.ph p • Open sourcing OCR technology for use in library settings: – http://minrvaproject.org/source.php 18
- 19. Sponsors • Institute of Museum and Library Services • University of Illinois Campus Research Board 19
- 20. Acknowledgements • My thanks to Ben Ryckman for Topic Space module development and support. • Many thanks to Chris Diaz, Residency Librarian, Scholarly Communications and Collections, University of Iowa for help with participant recruitment, observation, and interviewing support in the user studies • Thanks to Mayur Sadavarte, Graduate Student in Computer Science at the University of Illinois and Nate Ryckman, Graduate Student in Information Systems Management at Carnegie Mellon University for Optical Character recognition programming support. • Yinan Zhang, PhD Candidate in Computer Science at the University of Illinois, Sherry (Mengxue) Zheng, Graduate Student in Computer Science for help developing the search and suggestion functionality of the Deneb near-semantic index, Maria Lux, Graphic Designer for laying out the polished recommendations and prototyping Text-shot integration as a Minrva module. 20