Studies aggressive automated word removal in text categorization in large databases based on corpus statistics to reduce the noise in free texts and to enhance the computational efficiency of categorization. Topics include stop word identification, categorization methods for comparison, tests on four document collections, and evaluation…

Discussion of image database systems focuses on semantic queries and shows how an image is abstracted into a hierarchy of entity names and features; how relations are established between entities visible in the image; and how a "fuzzy" matching technique is used to compare semantic queries to image abstractions. (Author/LRW)

Discusses genetic algorithms in information retrieval, especially for relevance feedback, and evaluates the efficacy of a genetic algorithm with various order-based fitness functions for relevance feedback in a test database. Compares results with the Ide dec-hi method, one of the best traditional methods. (Contains 56 references.) (Author/LRW)

Compares the performance of inverted and signature file organizations in handling incrementally specified Boolean queries to an information retrieval system. The discussion covers the impact of more sophisticated signature file organizations, related problems, and possible future work. (65 references) (CLB)

Discusses end-user searching in Boolean information retrieval systems considers the role of search intermediaries and proposes a model of user preferences that incorporates a user's profile. Highlights include document representation; information queries; document output ranking; calculating user profiles; and selecting documents for a local…

Presents an improved inverted region quadtree variation called Dynamic Inverted Quadtree that is able to support and index a pictorial database. Pattern-searching algorithms are described, and the space needs of Fully Inverted and Dynamic Inverted Quadtrees are analytically compared based on a model of image randomness expressed as a branching…

Describes a procedure for information retrieval testing that is based on the comparison of statistically independent methods of retrieval applied to the same database. The probability ranking principle is discussed, the statistical meaning of relevance is examined, and the methodology is illustrated on a large database of MEDLINE records. (19…

Investigates the value of applying principles derived from simulated annealing to clustering tuples in database design, and compares this technique with a graph-collapsing clustering method. It is concluded that, while the new method does give superior results, the expense involved in algorithm run time is prohibitive. (24 references) (CLB)

Explores the evaluation of image and multimedia information-retrieval systems, particularly the effectiveness of several shape measures for content-based retrieval of similar images. Shape feature measures, or vectors, are computed automatically and can either be used for retrieval or stored in the database for future queries. (57 references)…

Discussion of relevancy in information retrieval systems focuses on an analysis of the efficiency of various cluster-based retrieval (CBR) strategies. A method for combining CBR and inverted index search is proposed that is cost effective in terms of time efficiency; and results of experiments are reported. (Contains 32 references.) (LRW)

Suggests a method that allows searchers to analytically compare the Boolean and probabilistic information retrieval approaches. Sample performance figures are provided for queries using the Boolean strategy, and for probabilistic systems. The variation of performance across sublanguages and queries is examined, as well as the performance of models…