SlugNERDS: A Named Entity Recognition Tool for Open Domain Dialogue Systems

SlugNERDS: A Named Entity Recognition Tool for Open Domain dialogue Systems Kevin K. Bowden, Jiaqi Wu, Shereen Oraby, Amita Misra, and Marilyn Walker Natural Language and dialogue Systems Lab University of California, Santa Cruz {kkbowden, jwu64, soraby, amisra2, mawalker}@ucsc.edu Abstract In dialogue systems, the tasks of named entity recognition (NER) and named entity linking (NEL) are vital preprocessing steps for understanding user intent, especially in open domain interaction where we cannot rely on domain-specific inference. UCSC’s effort as one of the funded teams in the 2017 Amazon Alexa Prize Contest has yielded Slugbot, an open domain social bot, aimed at casual conversation. We discovered several challenges specifically associated with both NER and NEL when building Slugbot, such as that the NE labels are too coarse-grained or the entity types are not linked to a useful ontology. Moreover, we have discovered that traditional approaches do not perform well in our context: even systems designed to operate on tweets or other social media data do not work well in dialogue systems. In this paper, we introduce Slugbot’s Named Entity Recognition for dialogue Systems (SlugNERDS), a NER and NEL tool which is optimized to address these issues. We describe two new resources that we are building as part of this work: SlugEntityDB and SchemaActuator. We believe these resources will be useful for the research community. Keywords: dialogue systems, entity recognition, knowledge graphs 1. Introduction When building dialogue systems, reliable named entity recognition (NER) and named entity linking (NEL) are vi- tal to understanding user intent, especially if these dialogue systems are open domain and intended to support conver- sations on any topic. In designing our open domain so- cial bot, Slugbot(Bowden et al., 2017b), for the 2017 Ama- zon Alexa Prize Contest(Ram et al., 2017), we discovered several challenges specifically associated with both NER and NEL. This paper discusses these challenges, and shows how we address them with Slugbot’s Named Entity Recog- nition for Dialogue Systems (SlugNERDS), a tool designed for NER and NEL in open domain dialogue. Additionally we present two corpora, SlugEntityDB and SchemaActua- tor, which are based on over 10,000 real user conversations with the system. We perform an extensive analysis of our system and the corpora to identify important areas of future work. NER and NEL have been actively researched topics for decades (Finkel and Manning, 2009; Ratinov and Roth, 2009; Ritter et al., 2011; Derczynski et al., 2015; Ni- tish Gupta and Roth, 2017). However, the resulting entity classification is often coarse and does not encode an on- tology. For example, Stanford NER features only a small number of abstract entity types such as PERSON, LO- CATION, ORGANIZATION, and MISC (Manning et al., 2014; Finkel and Manning, 2009); these categories don’t provide enough information for dialogue interpretation and generalization. Although other resources such as that from Ratinov and Roth (2009) utilize additional external knowl- edge by extracting 30 gazetteers from both the web and Wikipedia, the entity types are still not as varied as we need, and the …

Content truncated to 500 words