OSU-2: Generating Referring Expressions with a Maximum Entropy Classifier
نویسندگان
چکیده
Selection of natural-sounding referring expressions is useful in text generation and information summarization (Kan et al., 2001). We use discourse-level feature predicates in a maximum entropy classifier (Berger et al., 1996) with binary and n-class classification to select referring expressions from a list. We find that while mention-type n-class classification produces higher accuracy of type, binary classification of individual referring expressions helps to avoid use of awkward referring expressions.
منابع مشابه
Combining symbolic and corpus-based approaches for the generation of successful referring expressions
We present an approach to the generation of referring expressions (REs) which computes the unique RE that it predicts to be fastest for the hearer to resolve. The system operates by learning a maximum entropy model for referential success from a corpus and using the model’s weights as costs in a metric planning problem. Our system outperforms the baselines both on predicted RE success and on si...
متن کاملUDel: Generating Referring Expressions Guided by Psycholinguistc Findings
We present an approach to generating referring expressions in context utilizing feature selection informed by psycholinguistic research. Features suggested by studies on pronoun interpretation were used to train a classifier system which determined the most appropriate selection from a list of possible references. This application demonstrates one way to help bridge the gap between computationa...
متن کاملGenerating referring expressions containing quantifiers
Recent work on the Generation of Referring Expressions has increased the generating capability of algorithms in this area. This paper asks whether the models underlying these proposals can still be used if even more complex referring expressions are generated. To discuss this issue, we will investigate a variety of referring expressions that pose difficulties to current generation algorithms. I...
متن کاملThe Prevalence of Descriptive Referring Expressions in News and Narrative
Generating referring expressions is a key step in Natural Language Generation. Researchers have focused almost exclusively on generating distinctive referring expressions, that is, referring expressions that uniquely identify their intended referent. While undoubtedly one of their most important functions, referring expressions can be more than distinctive. In particular, descriptive referring ...
متن کاملSentence Extraction for Legal Text Summarisation
We describe a system for generating extractive summaries of texts in the legal domain, focusing on the relevance classifier, which determines which sentences are abstract-worthy. We experiment with naı̈ve Bayes and maximum entropy estimation toolkits and explore methods for selecting abstract-worthy sentences in rank order. Evaluation using standard accuracy measures and using correlation confir...
متن کامل