Challenges posed by web-based retrieval of scientific papers: Okapi participation in TIPS

Abstract

This paper reviews some of the past Okapi research and discusses the role of Okapi in the current TIPS project. The main purpose is to report new challenges faced by probabilistic text retrieval in the web environment and to indicate some of the solutions that are currently under investigation. In this context, extraction of indexing units from formatted document sources, user interface design, implementation of field searching and query expansion within the framework of probabilistic searching are discussed. The problem of maintaining session continuity in the web environment and a possible solution to this problem are outlined. Other challenges posed by the open nature of the web environment are also indicated. These include the difficulty of delimiting the boundaries of a search session and the potential of the web for collaborative information retrieval. A system for collaboratively filtering documents based on their contents is described in this connection. Issues surrounding the integration of Okapi with other pieces of software being developed for the TIPS project are also briefly discussed.

Terminological paraphrase extraction from scientific literature based on predicate argument tuples

“…To enhance the search functions of PubMed, the largest biomedical literature database in the world, Lu et al [10] introduced the automatic term mapping method, which automatically maps user queries into MeSH descriptors and enables QE with various types of thesaurus information. There have been many studies of QE application to improve the performance of biomedical information retrieval with controlled vocabularies such as MeSH and UMLS [6, 8, 11, 12, 37–39].…”

Exploring the Unexplored: Identifying Implicit and Indirect Descriptions of Biomedical Terminologies Based on Multifaceted Weighting Combinations

“…The above writing styles introduce some challenges in scholarly information retrieval (IR) [ 7 ] and information extraction (IE) from technical documents [ 8 ]. In particular, the queries of scholarly document retrieval are mostly technical terms [ 9 ] and thus it is very likely to miss the relevant documents not containing the query terms themselves but including only their meanings, which can negatively affect the ranking models by giving implicitly incorrect statistics as shown in Figure 1 .…”