A microcomputer-based information retrieval system with relational thesauri to support a medical expert system.
详细信息
- 作者：Wang ; Guang-Nay.
- 学历：Doctor
- 年：1989
- 导师：Evens, Martha W.
- 毕业院校：Illinois Institute of Technology
- 专业：Computer Science.;Information Science.;Artificial Intelligence.;Health Sciences, Medicine and Surgery.
- CBH：9027323
- Country：USA
- 语种：English
- FileSize：4852534
- Pages：128
文摘
LITREF is an online bibliographic retrieval system. It supports the Stroke Consultant MAIESTRO, an expert system for the diagnosis and management of stroke cases. This expert system is a joint project of the Department of Neurology at Michael Reese Hospital and the Department of Computer Science at Illinois Institute of Technology. LITREF interprets a Boolean command language to provide fast, efficient browsing of abstracts from the stroke literature. User feedback on keywords derived from a relational thesaurus is used to enhance the original query. Response time is 'immediate' when LITREF runs on an IBM PC/AT microcomputer. The user-friendly online help facilitates users access to the system. It has been enthusiastically received by users at its initial trial.;The system design uses an inverted file structure with a bitmap strategy to speed access to retrieval objects. The implementation uses MS-QUICK C under DOS 3.1. It runs on any IBM compatible PC with a hard disk. The database contains 597 abstracts in the area of Cerebral Infarction and Cerebral Hemorrhage from 1983 to 1988 and occupies 1.6 Megabytes in storage. Three index files take 42 Kbytes in memory. The cosine function is used to measure the similarity between a query and an abstract. Recall and precision are used in the evaluation of retrieval effectiveness.;Two experiments have been performed using LITREF as a testbed. One experiment involves applying alternative suffixing algorithms to both index terms and query keywords. When the results are evaluated by comparing precision values at each recall level we found that the word stem index method is superior to the full work index method. The other experiment is a test of whether a relational thesaurus extracted from a large lexicon with lexical and semantic relations from Webster's Seventh New Collegiate Dictionary can actually improve retrieval effectiveness. The initial test using binary weighting in a batch run obtained a negative result.