Douglas P. Metzler

Mental rotation: effects of dimensionality of objects and type of task

The original studies of mental rotation estimated rates of imagining rotations that were much slower when two simultaneously portrayed three-dimensional shapes were to be compared (R. Shepard & J. Metzler) than when one two-dimensional shape was to be compared with a previously learned two-dimensional shape (Cooper and her associates). In a 2 X 2 design, we orthogonally varied dimensionality of objects and type of task. Both factors affected reaction times. Type of task was the primary determiner of estimated rate of mental rotation, which was about three times higher for the single-stimulus task. Dimensionality primarily affected an additive component of all reaction times, suggesting that more initial encoding is required for three-dimensional shapes. In the absence of a satisfactory way of controlling stimulus complexity, the results are at least consistent with the proposal that once three-dimensional objects have been encoded, their rotation can be imagined as rapidly as the rotation of two-dimensional shapes.

Conjunction, ellipsis, and other discontinuous constituents in the constituent object parser

Abstract The Constituent Object Parser (COP) is a domain independent syntactic parser developed for use in information retrieval and similar applications. Its purpose is to extract a simple hierarchical description of a phrase or sentence that can be used in very general pattern matching procedures to determine the structural similarity of sentences or phrases that contain equivalent terms. This paper discusses the mechanisms by which COP handles the problems of conjunction, ellipsis, and discontinuous constituents. These structures are usually particularly difficult to handle in a parser that does not employ domain knowledge or even general semantic knowledge. cops mechanisms for these structures are directly tailored for, and, in part, even made possible by, the nature of the intended uses of the outputs by the information retrieval matching procedures.

Constituent object parsing for information retrieval and similar text processing problems

The architecture and functioning of the Constituent Object Parser are described. This system has been developed specifically for text processing applications such as information retrieval, which can benefit from structural comparisons between elements of text such as a query and a potentially relevant abstract. The general way in which the system performs these matches, and the ways in which this objective influenced the design of the system are described. The parsing architecture incorporates several interesting features including: (1) an unusual combination of declarative and procedural representation techniques, (2) a monotonic discipline which permits useful heuristic approaches to difficult linguistic problems such as ellipsis, conjunction, ambiguity, and ill‐formed and incomplete input, and (3) an attempt to minimize the level of syntactic detail required in both the grammar and the lexicon.

Hospital care watch (HCW): an ontology and rule-based intelligent patient management assistant

Hospital care watch (HCW) is a prototype medical intelligent assistant to improve patient care and safety and to reduce medical errors in the hospital setting. The system is based on an ontology of hospital care concepts including hospital activities, procedures, and policies, and insurance policies, as well as medical knowledge per se. A set of abstract rule types model general situations that need to be identified to detect potential problems and particular instantiations of those rule types identify individual situations. The system is not designed to provide medical diagnosis or other professional decision making; but rather to track the implications of such medical decisions taken by physicians and other medical professionals within the context of the guidelines and regulations of the medical environment, and the background of established medical knowledge.

The constituent object parser: syntactic structure matching for information retrieval

There has long been interest in the idea of using syntactic information as part ‘of an information retrieval strategy. People clearly gather information about the meaning of text (e.g., a sentence) both from the meanings of the individual words contained in it and from the structure in which the individual pieces sic put together. Since syntax reflects this structure, it would seem that using syntactic information in addition to information about the presence of query terms would increase the performance of an information retrieval system. Conventional information retrieval systems do not employ syntactic information. They are primarily term based. Thesauri or similar methods can be used to generalize or speciaiize the actual terms included in the query. But or~ly crude constraints on the organizational relationships between terms may be expressed, such as that they must appear in the same sentence, or within a given number of words of each other, or in a certain order. There is no way to express syntactic relationships, such as which term is the head of a phrase and which terms arc its modifiers or what the scope of a modifier is. For example, consider the terms junior and iollege. A query requesting that they both appear in the same sentence could retrieve documenis about junior college and college juniors as well as documents in which the two are barely related. (e.g., high school juniors attending summer classes at a local college). A user interested in documents about junior college may well have no interest in college juniors per se, and certainly does not want documents in which the two terms are not related at all, but that “accidentally” fit the query.

QUE: explanation through exploration

Abstract QUE, QUerying the Expert system, is designed to provide a collaborative exploratory environment for a variety of types of users of expert systems. Explanation in intelligent systems has long been the subject of research from a number of different approaches, each of which has shown promise as well as problems. This paper describes how QUE, which is currently under development, approaches the problem of providing explanation by providing tools for both developers and users that enable explanation and user understanding of the underlying expert systems reasoning.

The flexibility of case grammar representations: a porting procedure for natural language interfaces

Abstract While case grammars have frequently been used to implement natural language interfaces to various sorts of systems, there has been little agreement over what precisely the different cases ought to be, or exactly where the semantic boundaries between different cases ought to be drawn. As a result, the domain specific decisions taken in the implementation of a case based natural language interface can result in a system which is very difficult to modify to meet the demands of a different application domain. This paper describes an approach to analysing the case grammar structures used for particular domains and utilizing these analyses to facilitate the porting of case based natural language interfaces from existing domains to new ones. The analysis is based on the assumption that there is not a single correct set of cases that is applicable across all of language and world knowledge, but rather that a set of basic cases is transformed somewhat by the semantic requirements encountered in any particular domain. A type hierarchy of cases is developed consisting of three levels of generality. The lowest, most specific, level of the hierarchy consists of a domain specific mapping of the general cases dictated by the semantic requirements of the domain. The analysis procedure consists of explicating the mappings for the two domains and comparing them in order to discover the necessary shifts in representation required to capture the shifts in domain semantics.