The mission of NSFs National STEM Education Distributed Learning (NSDL) is to support development and coordination of educational resource collections. NSDL provides integrated access to the resulting network of learning resources through automated services. In addition to providing keyword search, NSDL organizes these educational resources using Strand maps. Each node in a Strand map represents a learning benchmark and these benchmarks are connected by links describing prerequisite relationships. The grade levels increase from bottom to top, as complex concepts taught in later grades build on simpler concepts taught in lower grade. The major topics that occur in the map are placed from left to right. By following paths on the Strand map, the users obtain access to related content (Figure 3). While Strand maps [4,13,169] and topic maps (such as NSDLs TM4L[7,12]), allow the user to vigorously engage in exploration activities [32,33,89,138], these are static and also known to place extraneous load on users . Thus, accessing internal and external resources of NSDL effectively requires a proper understanding of the personal activity context, context-aware resource discovery, and peer-network driven resource and knowledge sharing and collaborative recommendations. Lack of Personal Context-Aware Search and Navigation: For effective resource discovery, keyword search needs to be highly precise and informed of users search context. Let us consider two users who are searching for materials having to do with the keyword entropy. For the sake of the example, let us assume that the first user is a computer science teacher, who is interested in finding materials related to entropy within computer science context. Let us further assume that the second user is a physics student, who, naturally, interprets the term within its physics context. For these two users, the keywords in the documents in the collection as well as the keywords in their respective queries carry different meanings and connotations (Figure 4). When the contexts are taken into account, the search should rank and organize results differently for these two users. Lack of Peer-Networking and Recommendations: Consider John who is preparing a new course on bioinformatics with particular emphasis on information integration. While John can certainly query NSDL for resources on these topics, he will need to explore and weed out many irrelevant content before identifying a suitable set of materials. This is because content-based searches are not sufficiently precise. A future version of NSDL should help John in two complementary ways: (a) peer recommendation: Based on Johns interactions with the library (including his past queries, exploration emphasis he has declared, and the Strand maps he has explored), NSDL can help John locate other individuals (or peers) with similar interests or have used NSDL with similar goals; (b) resource recommendations based on peer networks: NSDL can recommend John resources that were explored and used by others within similar activity contexts. Lack of Content Previews and Exploration support: When searching NSDL, simple lists of documents returned as answers may not be effective. Instead, the system needs to provide snippets, summaries, and other preview techniques to help the user explore resources effectively. Consider a geography teacher who is searching for teaching materials on earthquakes. When this user is provided with a list of documents, she faces a number of challenges: First of all, each document in the list likely to contain multiple sub-topics and only a fraction of these will be relevant to the users context. Secondly, long documents increase the overhead on this user while sifting through the list. Instead the system should provide (a) context-aware ranking, (b) snippet extraction, (c) key concept highlighting, and (d) Strand map and resource summarization servic
|Effective start/end date||9/15/10 → 8/31/14|
- National Science Foundation (NSF): $509,168.00
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.