Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Demonstrations

Yang Liu, Tim Paek, Manasi PatwardhanNAACL 2018

TL;DR

This document presents the proceedings of the NAACL HLT 2018 Demonstrations Session, showcasing 20 cutting-edge natural language processing (NLP) systems. These demonstrations offer a glimpse into the practical applications and advancements in NLP, ranging from argument analysis and claim detection to medical scribes and social chatbots.

What Problem Does This Paper Address?

The NAACL HLT 2018 Demonstrations Session aims to bridge the gap between theoretical NLP research and real-world applications. It provides a platform for researchers and developers to showcase their systems and programs, allowing for direct interaction and feedback from the NLP community. This session addresses the need to demonstrate the practical utility and usability of NLP technologies, moving beyond purely theoretical contributions to tangible, working systems.

Methods

This document does not detail specific methods as it is a collection of demonstration abstracts. However, the listed demonstrations cover a wide array of NLP tasks and techniques, including:

  • Argumentation Mining: Identifying and analyzing arguments within text.
  • Claim Detection: Identifying verifiable claims in text.
  • Stance Detection: Determining the author's stance towards a target.
  • Sentiment Lexicon Debugging: Tools for improving sentiment analysis resources.
  • Cross-lingual Entity Extraction and Linking: Identifying and linking entities across different languages.
  • Entity Resolution and Disambiguation: Identifying and resolving mentions of entities in text, particularly in specialized domains.
  • Structural Ambiguity Games: Interactive tools for understanding linguistic ambiguity.
  • Vietnamese NLP Toolkit: A comprehensive toolkit for processing Vietnamese text.
  • Client-side NLP: Deploying NLP models directly in web browsers.
  • Medical Text Representation: Generating meaningful representations of medical texts.
  • Machine Translation Evaluation: Tools for visualizing and evaluating MT metrics.
  • Social Connection Analysis: Understanding social relationships from text.
  • Risk Detection in Financial Reports: Identifying risk-related information in financial documents.
  • Multi-modal Interaction: Combining language and gestures for AI embodiment.
  • Decision Conversation Analysis: Analyzing conversations related to decision-making.
  • Social Chatbots: Developing user-centric and content-driven chatbots.

Datasets

The paper does not explicitly list datasets used for training or evaluation for each demonstration. However, the nature of the demonstrations suggests the use of various publicly available and potentially proprietary datasets relevant to each specific NLP task.

Key Insights and Ideas

  • Practicality is Key: The sheer diversity of demonstrations highlights a strong trend towards building practical, deployable NLP systems that address real-world problems.
  • Cross-Lingual and Multi-modal are Growing: The presence of Vietnamese NLP tools and multi-modal interaction systems indicates progress in handling diverse languages and modalities.
  • Specialized Domains are Targeted: Demonstrations focusing on medical texts, financial reports, and ancient temple data show a move towards domain-specific NLP solutions.
  • Interactive Tools are Valuable: The inclusion of debugging tools, games, and visualization platforms emphasizes the importance of user interaction and understanding in NLP development.
  • Ethical and Social Implications: Demonstrations on social connections and risk detection touch upon the broader societal impact of NLP technologies.

Key Results

As this is a proceedings document for a demonstration session, specific quantitative results for each system are not provided in a consolidated manner. The success of each demonstration is implicitly measured by its acceptance into the program and its ability to engage and impress attendees. The document lists 20 accepted demonstrations out of 39 submitted papers, indicating a competitive selection process.

Limitations and Future Work

This document does not discuss limitations or future work for the collection as a whole. Each individual demonstration would have its own set of limitations and proposed future directions, which are not detailed in this overview. However, the existence of a dedicated demonstration session implies a continuous effort to push the boundaries of NLP and explore new avenues for research and application.

Summary updated

BibTeX
@inproceedings{liu2018proceedings,
  author = {Yang Liu and Tim Paek and Manasi Patwardhan},
  title = {{Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Demonstrations}},
  booktitle = {Conference of the North American Chapter of the Association for Computational Linguistics},
  year = {2018},
}
Figure