| In any online literature digital library, findability precedes usability: users cannot use what they cannot find. Four research directions that support better findability in digital libraries are (a) Accurate scoring functions to assign importance/prestige scores to publications, (b) Accurate similarity measures for publications to locate publications similar to a given publication, (c) Accurate ranking measures to order search results based on their importance and relevance to users' interests, and (d) Helping users develop search keywords that lead to successful searches.; The contributions of this thesis are as follows. (1) Propose and comparatively evaluate score functions for publications, authors, and publication venues, as well as similarity measures for publications, towards research direction items a and b. (2) Validate a new model for the evolution of research and citation behavior, namely, the Research Pyramid Model. Then, propose and evaluate two algorithms for identifying research pyramid structures in publication citation graphs, and for research-pyramid-based publication score generation, towards research direction item a. (3) Propose and evaluate a citation-based publication popularity growth and decay model, towards research direction item c. (4) Using the Research Pyramid Model and the identified research pyramid structures, develop two literature digital library searching and ranking tools: (a) A Research-Pyramid-based ranking tool that assigns accurate scores for publications, and (b) A scalable content-driven Search-Keyword Suggester that helps users to put together query search terms effectively. |
