This paper introduces a novel user interface that integrates search and browsing of very large category hierarchies with their associated text collections. A key component is the separate but simultaneous display of the representations of the categories and the retrieved documents. Another key component is the display ofmultiple selected categories simultaneously, complete with their hierarchical context. The prototype implementation uses animation and a three-dimensional graphical workspace to accommodate the category hierarchy and to store intermediate search results. Query specification in this 3D environment is accomplished via a novel method for painting Boolean queries over a combination of category labels and free text. Examples are shown on a collection of medical text.

We have developed an in6 vative searchin terface that allowsnAOz5z ert users to move through large in97z86 tion spacesin a flexible manle without feelin lost. The design goal was to o#er users a "browsin the shelves" experien5 seamlessly in tegrated with focused search. Key to achievin our goal is the explicit exposure of hierarchical faceted metadatain a manz6 that is in tuitive an in vitin to users. After several iteration of design an testinA the usability results are strikinA] positive. We believe our approach marks a major step forward in search userin terfacesan can serve as a model for web-based collection of up to 100,000 items.

We develop an automatic text categorization approach and investigate its application to text retrieval. The categorization approach is derived from a combination of a learning paradigm known as instancebased learning and an advanced document retrieval technique known as retrieval feedback. We demonstrate the effectiveness of our categorization approach using two real-world document collections from the MEDLINE database. Next we investigate the application of automatic categorization to text retrieval. Our experiments clearly indicate that automatic categorization improves the retrieval performance compared with no categorization. We also demonstrate that the retrieval performance using automatic categorization achieves the same retrieval quality as the performance using manual categorization. Furthermore, detailed analysis of the retrieval performance on each individual test query is provided. Index Terms: Text Categorization, Automatic Classification, Text Retrieval, Instance-Based Le...

Few studies have compared the interactive use of Boolean and natural language searching systems. We studied the use of three retrieval systems by senior medical students searching on queries generated by actual physicians in a clinical setting. The searchers were randomized to search on two of three different retrieval systems: a Boolean sys-tem, a word-based natural language system, and a con-cept-based natural language system. Our results showed no statistically significant differences in recall or precision among the three systems. Likewise, we found no user pref-erence for any system over the others. In the course of this study we did find, however, a number of problems with tra-ditional measures of retrieval evaluation when applied to the interactive search setting.

ABSTRACT: Full-text documents are a vital and rapidly growing part of online biomedical information. A single large document can contain as much information as a small database, but normally lacks the tight structure and consistent indexing of a database. Retrieval systems will often miss highly relevant parts of a document if the document as a whole appears irrelevant. Access to full-text information is further complicated by the need to search separately many disparate information resources. This research explores how these problems can be addressed by the combined use of two techniques: (1) natural language processing for automatic concept-based indexing of full text, and (2) methods for exploiting the structure and hierarchy of full-text documents. We describe methods for applying these techniques to a large collection of fulltext documents drawn from the Health Services/Technology Assessment Text (HSTAT) database at the National Library of Medicine (NLM), and examine how this hierarchical concept indexing can assist both document- and source-level retrieval in the context of NLM’s Information Sources Map project. Wright et al. 2

In this paper we describe an innovative query expansion evaluation framework (QEEF) which discovers the ontological and algorithmic characteristics that drive successful query expansion. The method consists of identifying UMLS (Unified Medical Language System) concepts in the Ohsumed corpus queries and documents, and then applying a variety of query expansion algorithms to the query concepts, both individually and at the query level. We analyse the results, discovering the characteristics of high relevance medical query expansions. We directly evaluate query expansion success, and this enables discovery of the relationship between the UMLS facets and this success. The paper details the methods used, and then discusses the influence of both UMLS attributes, and choice of query expansion algorithm, on query expansion success.

Abstract not found

An important challenge for biomedical information retrieval (IR) is dealing with the complex, inconsistent and ambiguous biomedical terminology. Frequently, a concept-based representation defined in terms of a domain-specific terminological resource is employed todeal with this challenge. In thispaper, weapproachtheincorporationofaconcept-based representation in monolingual biomedical IR from a crosslingual perspective. In the proposed framework, this is realized by translating and matching between text and conceptbased representations. The approach allows for deployment of a rich set of techniques proposed and evaluated in traditional cross-lingual IR. We compare six translation models and measure their effectiveness in the biomedical domain. We demonstrate that the approach can result in significant improvements in retrieval effectiveness over word-based retrieval. Moreover, we demonstrate increased effectiveness of a CLIR framework for monolingual biomedical IR if basic translations models are combined.

has a unique and leading position in the evolution of thesauri and metathesauri. Features that set it apart are: its composition from more than fifty component health care vocabularies; the sophisticated UMLS ontology linking the Metathesaurus with structures such as the Semantic Network and the SPECIALIST lexicon; and the high level of social collaboration invested in its construction and growth. It is our thesis that in order to successfully harness such a complex vocabulary for text retrieval we need sophisticated methods derived from a deeper understanding of the UMLS system. Thus we propose a theoretical framework based on the theory of rough sets, that supports the systematic and exploratory investigation of the UMLS Metathesaurus for text retrieval. Our goal is to make it more feasible for individuals such as patients and health care professionals to access relevant information at the point of need.

Conventional full-text systems represent documents as sets of index terms, and queries to these systems often retrieve irrelevant material when search terms occur in inappropriate contexts. We have developed document representations that capture the semantic contexts in which text words occur. Many bodies of literature contain stereotypic categories of information. For example, articles describing medical research consistently discuss interventions and outcomes. These semantic themes provide context for terms in the text, and thus, can facilitate precise fulltext searches. We have used a contextual model of clinical research articles, case reports, and review articles as the basis for a document representation in a full-text retrieval system.