Tanya Korelsky

Multidocument summarization via information extraction

We present and evaluate the initial version of RIPTIDES, a system that combines information extraction, extraction-based summarization, and natural language generation to support user-directed multidocument summarization.

On the need for domain communication knowledge

Much research in text planning has been oriented towards identifying the rhetorical component of communication knowledge that allows for the structuring of coherent multi‐sentential or even multi‐paragraph text. However, several researchers have recently found rhetoric‐based top‐down approaches to be inadequate for their particular needs. In this paper, we explore the issue by introducing the concept of “domain communication knowledge” as a new analytical tool. By analyzing two existing rhetoric‐based text planners, and by closely analyzing a particular genre, namely reports, we conclude that domain‐specific knowledge about communication is pervasive and should be represented explicitly for the text planning task.

Applied Text Generation

While we were able to exploit existing research for many of the design issues, it turned out that we needed to develop our own approach to text planning (Rambow 1990). This paper will present the system and a t tempt to show how these design objectives led to particular design decisions. The structure of the paper is as follows. In Section 2, we will present the underlying application and give examples of the output of the System. In Section 3, we will discuss the overall s tructure of Joyce. We then discuss the three main components in turn: the text planner in Section 4, the sentence planner in Section 5 and the realizer in Section 6. We will discuss the text planner in some detail since it represents a new approach to the problem. Section 7 traces the generation of a short text. In Section 8, we address the problem of portability, and wind up by discussing some shortcomings of Joyce in the conclusion.

Handling Stuctural Divergences and Recovering Dropped Arguments in a Korean/English Machine Translation System

This paper describes an approach for handling structural divergences and recovering dropped arguments in an implemented Korean to English machine translation system. The approach relies on canonical predicate-argument structures (or dependency structures), which provide a suitable pivot representation for the handling of structural divergences and the recovery of dropped arguments. It can also be converted to and from the interface representations of many off-the-shelf parsers and generators.

Inducing lexico-structural transfer rules from parsed Bi-texts

This paper describes a novel approach to inducing lexico-structural transfer rules from parsed bi-texts using syntactic pattern matching, statistical co-occurrence and error-driven filtering. We present initial evaluation results and discuss future directions.

Learning domain-specific transfer rules: an experiment with Korean to English translation

We describe the design of an MT system that employs transfer rules induced from parsed bitexts and present evaluation results. The system learns lexico-structural transfer rules using syntactic pattern matching, statistical co-occurrence and error-driven filtering. In an experiment with domain-specific Korean to English translation, the approach yielded substantial improvements over three baseline systems.

Knowledge requirements for the automatic generation of project management reports

The automatic generation of project management reports for software engineering projects puts special demands on the knowledge representation of software engineering environments (SEEs). This knowledge can best be represented as a process model, and must include a rich type hierarchy and various kinds of entity relations. Furthermore, histories must be maintained for certain types of information, and specific information about problems must be available. These types of knowledge, while useful in particular for automatic project reporting, are more generally important if a SEE is to provide comprehensive and useful services to a user.<<ETX>>

Security modeling in the Ulysses environment

The authors give an overview of how the Ulysses system can be used for security modeling. The default theory of security permits the security analysis of complex designs by decomposing them into their parts. System specifications may be made by using a specialized graphical language interface and a textual interface. In addition, there are a number of support tools which aid the modeler. One of these tools is the natural language component, which allows users to automatically generate short English descriptions of a graphical design. There is also a library component that facilitates the reuse of secure designs. In addition, there are mechanisms for adding security theories and other mathematical facts to augment the power of the theorem prover.<<ETX>>