Dependency Parsing for Spoken Dialog Systems

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 1513–1519, Hong Kong, China, November 3–7, 2019. c⃝2019 Association for Computational Linguistics 1513 Dependency Parsing for Spoken Dialog Systems Sam Davidson1, Dian Yu2, Zhou Yu2 1Department of Linguistics 2Department of Computer Science University of California, Davis Davis, CA 95616, USA {ssdavidson,dianyu,joyu}@ucdavis.edu Abstract Dependency parsing of conversational input can play an important role in language un- derstanding for dialog systems by identify- ing the relationships between entities extracted from user utterances. Additionally, effec- tive dependency parsing can elucidate dif- ferences in language structure and usage for discourse analysis of human-human versus human-machine dialogs. However, models trained on datasets based on news articles and web data do not perform well on spoken human-machine dialog, and currently avail- able annotation schemes do not adapt well to dialog data. Therefore, we propose the Spoken Conversation Universal Dependencies (SCUD) annotation scheme that extends the Universal Dependencies (UD) (Nivre et al., 2016) guidelines to spoken human-machine dialogs. We also provide ConvBank, a conver- sation dataset between humans and an open- domain conversational dialog system with SCUD annotation. Finally, to demonstrate the utility of the dataset, we train a dependency parser on the ConvBank dataset. We demon- strate that by pre-training a dependency parser on a set of larger public datasets and fine- tuning on ConvBank data, we achieved the best result, 85.05% unlabeled and 77.82% la- beled attachment accuracy. 1 Introduction Syntactic parsing (Chen et al., 2018) and seman- tic parsing (Kollar et al., 2018; Gupta et al., 2018) have been used in dialog systems to disambiguate the relationships between noun phrases in under- standing tasks. Compared to constituency pars- ing and semantic role labeling, dependency pars- ing provides more clear relationships between predicates and arguments (Johansson and Nugues, 2008). Constituency parsers provide information about noun phrases in a sentence, but provide only limited information about relationships within a noun phrase. For example, in the sentence “What do you think about Google’s privacy policy be- ing reviewed by journalists from CNN?,” a con- stituency parser would place “Google’s privacy policy being reviewed by journalists from CNN” under a single phrasal node. Similarly, a seman- tic role labeling system would tend to label the same phrase as an argument of the verb, but it would not disambiguate the relationships within the phrase. Finally, NER only provides informa- tion about named entities which may or may not be the key semantic content of the sentence. De- pendency parsers, by contrast, can provide infor- mation about relationships when a sentence con- tains multiple entities, even when those entities are within the same phrase. Identifying relationships between entities in a user utterance can help a dialog system formu- late a more appropriate response. For instance, in the sentence about “Google’s privacy policy” mentioned above, there are multiple entities for the system to consider. The system must determine the most important entity in the utterance in or- der …

Content truncated to 500 words