Susanne M. Humphrey

Using literature-based discovery to identify disease candidate genes

We present BITOLA, an interactive literature-based biomedical discovery support system. The goal of this system is to discover new, potentially meaningful relations between a given starting concept of interest and other concepts, by mining the bibliographic database MEDLINE. To make the system more suitable for disease candidate gene discovery and to decrease the number of candidate relations, we integrate background knowledge about the chromosomal location of the starting disease as well as the chromosomal location of the candidate genes from resources such as LocusLink and Human Genome Organization (HUGO). BITOLA can also be used as an alternative way of searching the MEDLINE database. The system is available at http://www.mf.uni-lj.si/bitola/.

A recent advance in the automatic indexing of the biomedical literature

The volume of biomedical literature has experienced explosive growth in recent years. This is reflected in the corresponding increase in the size of MEDLINE, the largest bibliographic database of biomedical citations. Indexers at the US National Library of Medicine (NLM) need efficient tools to help them accommodate the ensuing workload. After reviewing issues in the automatic assignment of Medical Subject Headings (MeSH terms) to biomedical text, we focus more specifically on the new subheading attachment feature for NLMs Medical Text Indexer (MTI). Natural Language Processing, statistical, and machine learning methods of producing automatic MeSH main heading/subheading pair recommendations were assessed independently and combined. The best combination achieves 48% precision and 30% recall. After validation by NLM indexers, a suitable combination of the methods presented in this paper was integrated into MTI as a subheading attachment feature producing MeSH indexing recommendations compliant with current state-of-the-art indexing practice.

Automatic indexing of documents from journal descriptors: a preliminary investigation

A new, fully automated approach for indexing documents is presented based on associating textwords in a training set of bibliographic citations with the indexing of journals. This journal-level indexing is in the form of a consistent, timely set of journal descriptors (JDs) indexing the individual journals themselves. This indexing is maintained in journal records in a serials authority database. The advantage of this novel approach is that the training set does not depend on previous manual indexing of hundreds of thousands of documents (i.e., any such indexing already in the training set is not used), but rather the relatively small intellectual effort of indexing at the journal level, usually a matter of a few thousand unique journals for which retrospective indexing to maintain consistency and currency may be feasible. If successful, JD indexing would provide topical categorization of documents outside the training set, i.e., journal articles, monographs, WEB documents, reports from the grey literature, etc., and therefore be applied in searching. Because JDs are quite general, corresponding to subject domains, their most probable use would be for improving or refining search results.

Knowledge-based indexing of the medical literature: the indexing aid project

This article describes the Indexing Aid Project for conducting research in the areas of knowledge representation and indexing for information retrieval in order to develop interactive knowledge-based systems for computer-assisted indexing of the periodical medical literature. The system uses an experimental frame-based knowledge representation language, FrameKit, implemented in Franz Lisp. The initial prototype is designed to interact with trained MEDLINE indexers who will be prompted to enter subject terms as slot values in filling in document-specific frame data structures that are derived from the knowledge-base frames. In addition, the automatic application of rules associated with the knowledge-base frames produces a set of Medical Subject Heading (MeSH) keyword indices to the document. Important features of the system are representation of explicit relationships through slots which express the relations; slot values, restrictions, and rules made available by inheritance through “is-a” hierarchies; slot values denoted by functions that retrieve values from other slots; and restrictions on slot values displayable during data entry.

Multiple approaches to fine-grained indexing of the biomedical literature.

The number of articles in the MEDLINE database is expected to increase tremendously in the coming years. To ensure that all these documents are indexed with continuing high quality, it is necessary to develop tools and methods that help the indexers in their daily task. We present three methods addressing a novel aspect of automatic indexing of the biomedical literature, namely producing MeSH main heading/subheading pair recommendations. The methods, (dictionary-based, post- processing rules and Natural Language Processing rules) are described and evaluated on a genetics-related corpus. The best overall performance is obtained for the subheading genetics (70% precision and 17% recall with post-processing rules, 48% precision and 37% recall with the dictionary-based method). Future work will address extending this work to all MeSH subheadings and a more thorough study of method combination.

MedIndEx system: medical indexing expert system

Abstract This report describes the MedIndEx System, a prototype for interactive knowledge-based indexing of the medical literature, being developed in an ongoing research project at the Lister Hill National Center for Biomedical Communications, National Library of Medicine. We first review current indexing practice, which, although performed interactively, relies heavily on published tools. We then discuss knowledge-based systems using frames. The prototype, which uses knowledge-base (KB) frames to guide indexers in completing indexing frames, defined as KB frames linked to specific documents, is described subsequently. It produces as output not only a set of indexing frames for each document, but also conventional MeSH index terms, some of which are generated by inferencing rules encoded in the KB frames. Next, the various types of assistance provided by the system are presented and explained with examples. The report concludes by summarizing the research activities and accomplishments to date and plans for continued development of the system.

Evaluation of interactive knowledge-based systems: overview and design for empirical testing

An overview of levels and approaches in the evaluation of knowledge‐based systems is presented. There is a need for empirical studies using objective criteria in advance of completing the technical evaluation of such systems. A methodology for this type of evaluation developed for a particular knowledge‐based indexing system is presented. It is suggested that the proposed study may serve as a model for the design of any evaluation in which the results of existing intellectual procedures are compared with results achieved when these procedures are aided by use of an appropriate expert system.

Indexing biomedical documents: From thesaural to knowledge-based retrieval systems

This position paper advocates interactive knowledge-based indexing of the National Library of Medicines (NLMs) MEDLINE database. Initially, it establishes that in the current setting concept indexing is needed and cannot be fully automated. Compatibility between conventional and knowledge-based indexing is then highlighted, followed by discussion of indexing as a cognitive process. The section on knowledge-based indexing systems describes how NLMs MedIndEx prototype addresses problems in conventional indexing, and includes the contention that constructing a knowledge base adapted from a conventional classified thesaurus and indexing scheme is not as daunting as it may seem. Extension of the indexing prototype to an intelligent search assistant illustrates use of the same knowledge base to integrate indexing and retrieval applications. The paper concludes with selections from the Milstead report on the state of the art of subject analysis for large multidisciplinary bibliographic databases, followed by suggested future directions for knowledge-based indexing.

A knowledge-base for retrieval evaluation

Automated bibliographic retrieval is a domain ripe for the study of artificial intelligence techniques. We are investigating the role of computerized knowledge in the classification and retrieval of documents and have four main points in this paper. The first claim is that library patrons will reliably rank documents to a query. The second claim is that a computer algorithm can reference a knowledge-base and rank documents to a query, as well as people do. The Medical Subject Headings Graph Structure (MeSHI is a knowledge-base used in retrieval of biomedical literature. We have an algorithm that references MeSH and assesses the degree of match between a MeSH-encoded document and query. This relevance algorithm depends on the “is-a” relations in MeSH and performs well. The medical field has several extensive, computerized “knowledge-bases”. By analogical reasoning these knowledge-bases can be used to add relationships or edges to MeSH. Our third hypothesis, that these additional edges would improve retrieval, was not supported. Our fourth set of experiments show that the new relationships, such as “surgicaltreatment-for” or “pathological-finding-of”, can be handled in special ways so that they contribute to the calculation of relevance.

Illustrated description of an interactive knowledge based indexing system

This report discusses the Indexing Aid Project for conducting research in interactive knowledge-based indexing of the medical literature. After providing an overview and background, we describe and illustrate the Indexing Aid System using an extended example, highlighting the knowledge-based capabilities of the system, namely, inheritance and internal retrieval, enforcement of restrictions, and other functions implemented by procedural attachments, which are characteristic of frame-based knowledge representation languages. A feature which generates reports for evaluating the system is also shown. The paper concludes with discussion of the research plan. The project is part of the Automated Classification and Retrieval Program at the Lister Hill National Center for Biomedical Communications, the research and development arm of the National Library of Medicine.