Generating Referring Expressions in Context: The GREC Task Evaluation Challenges
نویسندگان
چکیده
Until recently, referring expression generation (reg) research focused on the task of selecting the semantic content of definite mentions of listener-familiar discourse entities. In the grec research programme we have been interested in a version of the reg problem definition that is (i) grounded within discourse context, (ii) embedded within an application context, and (iii) informed by naturally occurring data. This paper provides an overview of our aims and motivations in this research programme, the data resources we have built, and the first three sharedtask challenges, grec-msr’08, grec-msr’09 and grec-neg’09, we have run based on the data.
منابع مشابه
UDel: Named Entity Recognition and Reference Regeneration from Surface Text
This report describes the methods and results of a system developed for the GREC Named Entity Recognition and GREC Named Entity Regeneration Challenges 2010. We explain our process of automatically annotating surface text, as well as how we use this output to select improved referring expressions for named entities.
متن کاملProducing Referring Expressions in Dialogue: Insights from a Translation Exercise
This paper discusses some of the challenges which arise in comparing the outputs of an algorithm for generating referring expressions across languages and cultures. The context in which the algorithm is employed and evaluated is a virtual environment, and the referring expressions in question form part of a dialogue script “acted out” by two virtual agents in a furniture shop. This setup was cr...
متن کاملIS-G: The Comparison of Different Learning Techniques for the Selection of the Main Subject References
The GREC task of the Referring Expression Generation Challenge 2008 is to select appropriate references to the main subject in given texts. This means to select the correct type of the referring expressions such as name, pronoun, common, or elision (empty). We employ for the selection different learning techniques with the aim to find the most appropriate one for the task and the used attribute...
متن کاملThe TUNA-REG Challenge 2009: Overview and Evaluation Results
The TUNA-REG’09 Challenge was one of the shared-task evaluation competitions at Generation Challenges 2009. TUNAREG’09 used data from the TUNA Corpus of paired representations of entities and human-authored referring expressions. The shared task was to create systems that generate referring expressions for entities given representations of sets of entities and their properties. Four teams submi...
متن کاملThe UMUS System for Named Entity Generation at GREC 2010
We present the UMUS (Université du Maine/Universität Stuttgart) submission for the NEG task at GREC’10. We refined and tuned our 2009 system but we still rely on predicting generic labels and then choosing from the list of expressions that match those labels. We handled recursive expressions with care by generating specific labels for all the possible embeddings. The resulting system performs a...
متن کامل