This is a very short paper that briefly discusses some of the tasks NLG systems perform. It no research interest, but I have occasionally found it useful as way introducing to potential project collaborators who know nothing about field.

We examine the problem of generating definite noun phrases that are appropriate referring expressions; is, (a) successfully identify intended referent to hearer whilst (b) not conveying him or her any false conversational implicatures (Grice, 1975). review several possible computational interpretations implicature maxims, with different costs, and argue simplest may be best, because it seems closest what human speakers do. describe our recommended algorithm in detail, along a specification...

In this article, we give an overview of Natural Language Generation (NLG) from applied system-building perspective. The article includes a discussion when NLG techniques should be used; suggestions for carrying out requirements analyses; and description the basic tasks content determination, discourse planning, sentence aggregation, lexicalization, referring expression generation, linguistic realisation. Throughout, emphasis is on established that can used to build simple but practical...

This paper describes SimpleNLG, a realisation engine for English which aims to provide simple and robust interfaces generate syntactic structures linearise them. The library is also flexible in allowing the use of mixed (canned non-canned) representations.

I present an architecture for data-to-text systems, that is NLG systems which produce texts from non-linguistic input data; this essentially extends the of Reiter and Dale (2000) to whose raw data instead AI knowledge bases. This being used in BabyTalk project, based on experiences several projects at Aberdeen; it also seems be compatible with many developed elsewhere. It consists four stages are organised a pipeline: Signal Analysis, Data Interpretation, Document Planning, Microplanning Realisation.

There is growing interest in using automatically computed corpus-based evaluation metrics to evaluate Natural Language Generation (NLG) systems, because these are often considerably cheaper than the human-based evaluations which have traditionally been used NLG. We review previous work on NLG and validation of automatic NLP, then present results two studies how well some popular other areas NLP (notably BLEU ROUGE) correlate with human judgments domain computer-generated weather forecasts....

I survey some recent applications-oriented NL generation systems, and claim that despite very different theoretical backgrounds, these systems have a remarkably similar architecture in terms of the modules they divide process into, computations perform, way interact with each other. also compare this 'consensus architecture' among applied NLG psycholinguistic knowledge about how humans speak, argue at least aspects consensus seem to be agreement what is known human language production, fact...

Contemporary Neonatal Intensive Care Units collect vast amounts of patient data in various formats, making efficient processing information by medical professionals difficult. Moreover, different stakeholders the neonatal scenario, which includ

We simplify previous work in the development of algorithms for generation referring expressions while at same time taking account psycholinguistic findings and transcript data. The result is a straightforward algorithm that computationally tractable, sensitive to preferences human users, reasonably domain-independent. provide specification resources host system must order make use algorithm, describe an implementation used IDAS system.

Natural Language Generation (NLG) can be used to generate textual summaries of numeric data sets. In this paper we develop an architecture for generating short (a few sentences) large (100KB or more) time-series The integrates pattern recognition, abstraction, selection the most significant patterns, microplanning (especially aggregation), and realisation. We also describe evaluate SumTime-Turbine, a prototype system which uses textualsummaries sensor from gas turbines.

This article presents research conducted to establish how information is shared across the personal social network in sensitive context of a health crisis. We worked with parents very sick babies who were cared for hospital's Neonatal Unit (NNU). Through combination interviews, focus group, and surveys, we developed user model that wanted share, they adapted this individual recipients. then prototype software tool which created adaptive updates members parents' network. The contained...

We report our efforts in identifying a set of previous human evaluations NLP that would be suitable for coordinated study examining what makes more/less reproducible. present results and findings, which include just 13% papers had (i) sufficiently low barriers to reproduction, (ii) enough obtainable information, considered all but one the experiments we selected reproduction was discovered have flaws made meaningfulness conducting questionable. As result, change design from reproduce...