:: Brook Wu ::

Grants: Digital Library Research at NJIT

»   Highlight
»   IntegraL
»   EZ Search

(Note: This page is a duplicate of our handout distributed at the 2004 NSDL All Projects Meeting; it is made by Dr. Bieber.)

At NJIT, we’re focusing on the following aspects of Digital Library research:

§ integrating collections and services
§ automatically generating links to resources in related collections and relevant services
§ metainformation: the full description and context surrounding an element of interest
§ federated metasearch among heterogeneous search services
§ conceptual clustering of search results (see http://highlight.njit.edu/)
§ customizing links to the user's task, preferences and characteristics
§ recommender systems based on collaborative filtering, content-based and knowledge based recommendations
§ educational strategies for learning through digital libraries
§ rigorous evaluation of interaction, effectiveness and efficiency

We are proud to participate in the National Science Digital Library community (http://www.nsdl.org/) which provides a major series of testbeds, as well as much of our funding.

We are especially interested in our research being widely used. We are actively seeking collaboration with digital libraries, “brick and mortar” public and university libraries, and library database and service providers, as well as educators for developing effective learning strategies for digital library materials.

We've been fortunate to receive four federal grants, which are enabling us to support many undergraduate, masters and Ph.D. students in digital library research.

Digital Library Service Integration Grant (DLSI)

Screenshot of our current DLSI prototype, integrating two independent digital library systems: NASA’s National Space Science Data Center (NSSDC) Master Catalog and the Arizona Document Summarizer from the University of Arizona. NSSDC users query the database from a form, returning the screen above. DLSI added link anchors to elements it recognized: documents and launch dates (indicated by the circled "i" in the 2nd and 3rd columns). When the user selects a document anchor, DLSI generates the list of links shown. The first will prompt the NSSDC system to display this document. The second will prompt DLSI to pass the document to the external AI Summarizer system. Clicking on a launch date generates

a separate list of links to services relevant to that kind of element.

Dates: 9/02 - 8/05
Sponsor: National Science Foundation - National Science Digital Library Program
Investigators: Michael Bieber (PI), Il Im, Yi-Fang Wu (NJIT, Information Systems)
Partners: AskNSDL (University of Syracuse), Atmospheric Visualization Collection (Argonne National Laboratory), Metis
(University of Colorado, Boulder), NASA National Space Science Data Center
Project Demo: http://is.njit.edu/dlsi/dlsi-demo.htm

The Digital Library Service Integration (DLSI) project provides a systematic lightweight approach for integrating digital library collections and services. When the user clicks on an item within a digital library collection or service, DLSI automatically generates a set of links to related information and relevant services. The set of links is customized using collaborative filtering techniques, matching the current user's navigation to the "click streams" of other users.

Figure for the IntLib/IntegraL projects: A “mock up” of an IntLib/IntegraL integration with EBSCOhost. IntLib/IntegraL has automatically added link anchors (the blue “i” icons) to several elements of a search result. The two pop-up windows with links were generated automatically when the user clicked on the document title and on the key phrase “physics teaching” respectively. Selecting any link will send the appropriate request to its corresponding collection or service. Except for the metadata link, all links shown lead to collections and services within the National Science Foundation’s National Science Digital Library (NSDL) system.

IntLib Grant

Dates: 1/05 - 12/07
Sponsor: Institute of Museum and Library Services - National Leadership Grant for Libraries
Investigators: Michael Bieber (PI, NJIT IS), Luis Hernandez (Newark Public Library), Il Im (NJIT IS), Richard Sweeney (NJIT, Van Houten Library), Yi-Fang Wu (NJIT IS)
Partners: Newark Public Library, New Jersey Digital Highway, NJIT IS Department, NJIT Van Houten Library

The IntLib Grant focuses on integrating the resources of public libraries primarily (and university libraries secondarily) together with digital libraries. It builds upon the DLSI project. We plan to integrate selected resources within:

EBSCOhost (at Newark Public Library and NJIT's Van Houten Library)
Gale's Discovery Collection (at Newark Public Library)
On-line Catalog Systems (at Newark Public Library and NJIT's Van Houten Library)
ProQuest (at NJIT's Van Houten Library)
New Jersey Digital Highway

The IntLib software will be freely available to public libraries and digital libraries to integrate their collections and services (if desired and when permissible, with those of our partners).

IntegraL Grant

Dates: 12/04 - 11/07
Sponsor: National Science Foundation - National Science Digital Library Program
Investigators: Michael Bieber (PI, NJIT IS), Il Im (NJIT, IS), Richard Sweeney (NJIT, Van Houten Library), Yi-Fang Wu (NJIT, IS)
Partners: Cumberland County College, Ramapo College, Olin College of Engineering, JerseyClicks, Digital Library for Earth Science Education (DLESE), StartingPoint, Science@NASA, NJIT IS Department, NJIT Van Houten Library

The IntegraL Grant focuses on integrating specific resources of university libraries with those of the National Science Digital Library. It builds upon the DLSI project. We plan to integrate selected resources within:

• ACM Digital Library	• Digital Library for Earth Science Education (DLESE)
• Elsevier Science Direct (permission pending)	• Science@NASA
• NJIT Electronic Thesis collection	• NSDL Core Integration features
• JerseyClicks	• an on-line bookstore
• StartingPoint

The funding for the IntegraL and IntLib grants combined will enable us to develop two additional aspects that both projects will then use: a next generation collaborative filtering engine and next generation federated search. (IntLib is funding students and IntegraL is funding faculty to work together on these.)

The IntegraL software will be freely available to university libraries and members of the National Science Digital Library to integrate their collections and services (if desired and when permissible, with those of our partners).

General Recommendation Engine Grant (GRE)

Dates: 10/04 - 9/07
Sponsor: National Science Foundation - National Science Digital Library Program
Investigators: Yi-fang Brook Wu(PI, NJIT IS), Michael Bieber (NJIT IS), Vincent Oria (NJIT, Computer Science), Il Im (Yonsei University)
Partners: Digital Library for Earth Science Education (DLESE), Digital Library Collection for Computer Vision Education (Swarthmore College), EconPort Digital Library for Microeconomics Education (University of Arizona), Eisenhower National Clearinghouse for Mathematics and Science Education, NJIT IS Department, NJIT Van Houten Library

The General Recommendation Engine (GRE) project will develop the next generation of recommender systems, and apply these within the National Science Digital Library. GRE will supplement user's searches with sets of links that others have found useful. GRE will develop and combine three next generation approaches to recommendations: collaborative-filtering, content-based recommendations and knowledge-based recommendations to craft the best set of links to related information.

For collection builders and service providers, this project will provide a ‘plug-and-play’ type of recommendation service that the NSDL system developers (collection builders) to make various levels of integration with GRE possible – from a simple ‘Similar Documents’ link to more sophisticated, fully enumerated recommendations. The figure below shows mock-up screens of simple and more sophisticated recommendations by GRE.

Mockup screen of the ‘Similar Documents’ feature:

Clicking on the ‘Similar Documents’ link will display a list of documents similar to the current document (‘The Roma Personal Metadata Service’).

Mockup screen of the full recommendation feature

This recommendation service requires more information such as user information, URLs that users visited (clickstreams), and actual document contents. This information will be sent to GRE as users use the system. The recommendation list can be displayed whenever a user logs in or only when the user requests it (as in the example shown). In the example, if the user clicks on the ‘recommendations’ link, a list of recommended documents will be displayed.

Note that this differs from IntLib and IntegraL in that GRE's recommendations apply to user searching. IntLib and IntegraL's links are placed within regular content. GRE will use DLSI's automated linking as a fourth kind of recommendation on top of specific search items (treating these as regular content).