Philip J. Hayes

Steps toward graceful interaction in spoken and written man-machine communication

Natural language processing is often seen as a way to provide easy-to-use and flexible interfaces to interactive computer systems. White natural language interfaces typically perform well in response to straightforward requests and questions within their domain of discourse, they often fail to interact gracefully with their users in less predictable circumstances. Most current systems cannot, for instance: respond reasonably to input not conforming to a rigid grammar; ask for and understand clarification if their users input is unclear; offer clarification of their own output if the user asks for it; or interact to resolve any ambiguities that may arise when the user attempts to describe things to the system. We believe that graceful interaction in these and the many other contingencies that can arise in human conversation is essential if interfaces are ever to appear co-operative and helpful, and hence be suitable for the casual or naive user, and more habitable for the experienced user. In this paper, we attempt to outline key components of graceful interaction, to identify major problems involved in realizing them, and in some cases to suggest the shape of solutions. To this end we propose a decomposition of graceful interaction into a number of relatively independent skills: skills involved in parsing elliptical, fragmented, and otherwise ungrammatical input; in ensuring robust communication; in explaining abilities and limitations, actions and the motives behind them; in keeping track of the focus of attention of a dialogue; in identifying things from descriptions, even if ambiguous or unsatisfiable; and in describing things in terms appropriate for the context. We suggest these skills are necessary for graceful interaction in general and form a good working basis for graceful interaction in a certain large class of application domains, which we define. None of these components appear individually much beyond the current state of the art, at least for suitably restricted domains of discourse. Thus, we advocate research into the construction of gracefully interacting systems as an activity likely to pay major dividends in improved man-machine communication in a relatively short time.

Breaking the Man-Machine Communication Barrier

How can interactive systems be made more helpful and responsive? This tool-independent system cuts down frustration by incorporating many of the communications shortcuts people use naturally.

Parsing spoken language: a semantic caseframe approach

Parsing spoken input introduces serious problems not present in parsing typed natural language. In particular, indeterminacies and inaccuracies of acoustic recognition must be handled in an integral manner. Many techniques for parsing typed natural language do not adapt well to these extra demands. This paper describes an extension of semantic caseframe parsing to restricted-domain spoken input. The semantic caseframe grammar representation is the same as that used for earlier work on robust parsing of typed input. Due to the uncertainty inherent in speech recognition, the caseframe grammar is applied in a quite different way, emphasizing island growing from caseframe headers. This radical change in application is possible due to the high degree of abstraction in the caseframe representation. The approach presented was tested successfully in a preliminary implementation.

Graceful interaction through the COUSIN command interface

Currently available interactive command interfaces often fail to provide adequate error correction or on-line help facilities, leading to the perception of an unfriendly interface and consequent frustration and reduced productivity on the part of the user. The COUSIN project of Carnegie–Mellon University is developing command interfaces which appear more friendly and supportive to their users, using a form-based model of communication, and incorporating error correction and on-line help. Because of the time and effort involved in constructing truly user-friendly interfaces, we are working on interface system designed to provide interfaces to many different applica-tion systems, as opposed to separate interfaces to individual applications. A COUSIN interface system gets the information it needs to provide these services for a given application from a declarative description of that applications communication needs.

Flexible Parsing

When people use natural language in natural settings, they often use it ungrammatically, missing out or repeating words, breaking-off and restarting, speaking in fragments, etc., Their human listeners are usually able to cope with these deviations with little difficulty. If a computer system wishes to accept natural language input from its users on a routine basis, it must display a similar indifference. In this paper, we outline a set of parsing flexibilities that such a system should provide. We go on to describe FlexP. a bottom-up pattern-matching parser that we have designed and implemented to provide these flexibilities for restricted natural language input to a limited-domain computer system.

Classifying users: a hard look at some controversial issues

It has become a common recommendation to computer interface designers that they should “Know the User” (e.g., Rubinstein and Hersh, 1984). This panel discussion will examine the issues that arise from this advice.

Entity-Oriented Parsing

An entity-oriented approach to restricted-domain parsing is proposed. In this approach, the definitions of the structure and surface representation of domain entities are grouped together. Like semantic grammar, this allows easy exploitation of limited domain semantics. In addition, it facilitates fragmentary recognition and the use of multiple parsing strategies, and so is particularly useful for robust recognition of extragrammatical input. Several advantages from the point of view of language definition are also noted. Representative samples from an entity-oriented language definition are presented, along with a control structure for an entity-oriented parser, some parsing strategies that use the control structure, and worked examples of parses. A parser incorporating the control structure and the parsing strategies is currently under implementation.

Robust Parsing Using Multiple Construction-Specific Strategies

Robust natural language interpretation requires strong semantic domain models, “fail-soft” recovery heuristics, flexible control structures, and focused user interaction when automatic correction proves unfeasible. Although single-strategy parsers have met with some success, a multi-strategy approach, with strategies selected dynamically according to the type of construction being parsed at any given time, is shown to provide a higher degree of flexibility, redundancy, and ability to bring task-specific domain knowledge (in addition to general linguistic knowledge) to bear on both grammatical and ungrammatical input. This construction-specific, multi-strategy approach can also help provide tightly focused interaction with the user in cases of semantic or structural ambiguity by allowing such ambiguities to be represented without duplication of unambiguous material. The approach also aids in task-specific language development by allowing direct interpretation of languages defined in terms natural to the task domain, and with the definition of data base interfaces by facilitating a unified treatment of update and access requests. Two small experimental parsers that were implemented to illustrate these advantages are presented, followed by the description of a parsing algorithm that integrates several of the best features of the two smaller parsers, including case-frame instantiation and partial pattern-matching strategies. The algorithm can deal with conjunctions, fragmentary input, and ungrammatical structures, as well as less exotic, grammatically correct input.

A Construction-Specific approach to Focused Interaction in Flexible Parsing

A flexible parser can deal with input that deviates from its grammar, in addition to input that conforms to it. Ideally, such a parser will correct the deviant input; sometimes, it will be unable to correct it at all; at other times, correction will be possible, but only to within a range of ambiguous possibilities. This paper is concerned with such ambiguous situations, and with making it as easy as possible for the ambiguity to be resolved through consultation with the user of the parser - we presume interactive use. We show the importance of asking the user for clarification in as focused a way as possible. Focused interaction of this kind is facilitated by a construction-specific approach to flexible parsing, with specialized parsing techniques for each type of construction, and specialized ambiguity representations for each type of ambiguity that a particular construction can give rise to. A construction-specific approach also aids in task-specific language development by allowing a language definition that is natural in terms of the task domain to be interpreted directly without compilation into a uniform grammar formalism, thus greatly speeding the testing of changes to the language definition.

Uniform help facilities for a cooperative user interface

This paper describes the design of the help and explanation component of a user-friendly operating system command interface called COUSIN. The facility can provide two kinds of information: (1) static descriptions of the various subsystems that can be invoked, their parameters, and the syntax that must be used; (2) dynamically generated descriptions of the state of the current interaction, why that state has arisen, and what the users options for action are. Both types of information are presented in the same way through a network of small text frames connected by semantically motivated links in the style of the ZOG system. Frames containing static information are generated automatically for each subsystem from a declarative description of the subsystem which is also used by COUSIN for its other services, including spelling and grammar correction and interactive error resolution. Dynamically generated frames are incorporated temporarily into the static network with semantic links appropriate to the current command context.