Colleen Crangle

Language and Learning for Robots

From the Publisher: Robot technology will find wide-scale use only when a robotic device can be given commands and taught new tasks in a natural language. This text pursues a theoretical approach to language and learning in developing the notion of an instructable robot that derives intelligence from human interaction.

Collaborative conversational interfaces

This paper proposes a method of designing human-computer speech interfaces based on principles of human conversation. It argues that conversation is the primary mode of language use and that it is fundamentally collaborative. Speech interfaces should therefore be designed to recreate the collaborative nature of natural conversations. The paper presents five strategies for designingcollaborative conversational interfaces, and it describes the principles of human-language use that underly these strategies. The paper also argues that collaborative conversational interfaces have a crucial advantage over other kinds of interfaces in that they are readily adaptive to different levels of experience and styles of use. The paper gives examples of collaborative conversational interfaces that we have developed, and discusses the ways in which these interfaces have been made adaptive.

Mining experimental evidence of molecular function claims from the literature

MOTIVATION The rate at which gene-related findings appear in the scientific literature makes it difficult if not impossible for biomedical scientists to keep fully informed and up to date. The importance of these findings argues for the development of automated methods that can find, extract and summarize this information. This article reports on methods for determining the molecular function claims that are being made in a scientific article, specifically those that are backed by experimental evidence. RESULTS The most significant result is that for molecular function claims based on direct assays, our methods achieved recall of 70.7% and precision of 65.7%. Furthermore, our methods correctly identified in the text 44.6% of the specific molecular function claims backed up by direct assays, but with a precision of only 0.92%, a disappointing outcome that led to an examination of the different kinds of errors. These results were based on an analysis of 1823 articles from the literature of Saccharomyces cerevisiae (budding yeast). AVAILABILITY The annotation files for S.cerevisiae are available from ftp://genome-ftp.stanford.edu/pub/yeast/data_download/literature_curation/gene_association.sgd.gz. The draft protocol vocabulary is available by request from the first author.

Conversational interfaces to robots

There is growing interest in robots that are designed specifically to interact with people and which respond to voice commands. Very little attention has been paid, however, to the kind of verbal interaction that is possible or desirable with robots. This paper presents recent work in multimodal interfaces that addresses this question. It proposes a new form of robot-user interface, namely a collaborative conversational interface. This article explains what collaborative conversational interfaces are, argues for their application in robots, and presents strategies for designing good conversational interfaces. It concludes with a discussion of the particular challenges faced in designing conversational interfaces for robots.

Knowledge architectures for patient access to breast-cancer information

To provide resourceful information about breast-cancer diagnosis and treatment, the authors worked with frequently-asked question (FAQ) files to create an effective breast-cancer knowledge server. They discuss the challenge of using FAQs to accommodate breast-cancer patients and their diverse information needs.

Context-fixing semantics for instructable robots

Abstract Instructable robots must be able to interpret a wide range of ordinary natural-language commands. This paper presents an approach to the interpretation of commands that takes into account the context in which the commands are given. It shows how the precise interpretation of many ordinary English words can be fixed only within their context of use and not before. It examines the role of the perceptual situation in fixing that interpretation, the role of the cognitive and perceptual functioning of the robot, and the role of the immediate linguistic surround. Our approach draws on the model-theoretic tradition in semantics in that it defines a set of models in terms of which the English commands to the robot are interpreted. At the same time, it uses a procedural semantics for the lexicon, thus addressing the question of how the robot can use the instruction it is given to perform the task described by that instruction. Examples are drawn primarily from instruction in elementary mathematics, Other examples come from our recent work with a robotic aid for the physically disabled.†

Structural Similarities between Brain and Linguistic Data Provide Evidence of Semantic Relations in the Brain

This paper presents a new method of analysis by which structural similarities between brain data and linguistic data can be assessed at the semantic level. It shows how to measure the strength of these structural similarities and so determine the relatively better fit of the brain data with one semantic model over another. The first model is derived from WordNet, a lexical database of English compiled by language experts. The second is given by the corpus-based statistical technique of latent semantic analysis (LSA), which detects relations between words that are latent or hidden in text. The brain data are drawn from experiments in which statements about the geography of Europe were presented auditorily to participants who were asked to determine their truth or falsity while electroencephalographic (EEG) recordings were made. The theoretical framework for the analysis of the brain and semantic data derives from axiomatizations of theories such as the theory of differences in utility preference. Using brain-data samples from individual trials time-locked to the presentation of each word, ordinal relations of similarity differences are computed for the brain data and for the linguistic data. In each case those relations that are invariant with respect to the brain and linguistic data, and are correlated with sufficient statistical strength, amount to structural similarities between the brain and linguistic data. Results show that many more statistically significant structural similarities can be found between the brain data and the WordNet-derived data than the LSA-derived data. The work reported here is placed within the context of other recent studies of semantics and the brain. The main contribution of this paper is the new method it presents for the study of semantics and the brain and the focus it permits on networks of relations detected in brain data and represented by a semantic model.

WHAT WORDS MEAN: SOME CONSIDERATIONS FROM THE THEORY OF DEFINITION IN LOGIC

For those wanting to understand the capacity humans have to use language with perspicuity and power, the question of what words mean is a central one. The choice of just the right word at just the right place, in anything from a legal argument to a poem, greatly influences the overall effect, indeed the meaning, of what is said or written. Much time and effort has been spent in lexical semantics trying to provide justification for one particular view of the lexicon over the other. In all these discussions, however, one glaring omission Stands out. No evaluation of lexical theories has taken serious note of the theory of definition in logic. The purpose of this paper is to introduce into the discussion of lexical meaning considerations from this theory. I aim to show how these considerations produce scepticism about ever formulating one theoretically adequate, generalized description of what words mean.

AN ANALYSIS OF ‘IF’ SENTENCES FROM THE STANDPOINT OF COMMUNICATION

This article examines ‘if‘ sentences and asks how one can explain the acceptability of one sentence over another in specific written contexts. It takes into account the temporal relations expressed in ‘if‘ sentences and the communicative intents that lie behind their use. The analysis presented has its roots in our work on robots that understand English. We hold the strong conviction that the many subtle but clear distinctions that exist in English, such as those expressed in ‘if‘ sentences, are there for good purpose. They have developed historically and are used for effective and accurate communication. Although it may well be that our ultimate objective in communication with robots is to have the same completely free and sloppy use of language characteristic of spoken speech, a necessary first step toward this goal is to have robots understand the standard uses of English typically found in well-written text. It is in this framework that the analysis presented in this paper is conducted. Why are ‘if‘ sentences such a challenge in general to semantics? One reason is the wide range of complex notions they can be used to communicate. A recent issue of The Nm York Times Book Review (July 26, 1992) contains these examples. Notice not only what complex idea is being communicated in each case, but what specific temporal, logical, or causal relation is being described.