Crowdsourcing a Normative Natural Language Dataset: A Comparison of Amazon Mechanical Turk and In-Lab Data Collection

The current work seeks to characterize a unique genre of music, elevator music (Muzak), using behavioral crowd-sourcing data from Amazon Mechanical Turk. Participants rated excerpts of elevator music along with more rewarding genres of music for pleasure, valence, familiarity, and recognition. Our results demonstrate that elevator music is rated as neutrally pleasurable, with high valence, and highly familiar. Data collection is ongoing, and future experiments use computational models of music to tease apart the neutral effects of elevator music listening. Our results have practical significance in that they may provide a potential control musical stimulus to be used with self-selected rewarding music.

Download Full-text

Does hotel cleanliness correlate with surfaces guests contact?

International Journal of Contemporary Hospitality Management ◽

10.1108/ijchm-02-2018-0105 ◽

2019 ◽

Vol 31 (7) ◽

pp. 2933-2950 ◽

Cited By ~ 2

Author(s):

Haeik Park ◽

Sheryl Fried Kline ◽

Jooho Kim ◽

Barbara Almanza ◽

Jing Ma

Keyword(s):

Data Collection ◽

Empirical Data ◽

Design Methodology ◽

Mechanical Turk ◽

Amazon Mechanical Turk ◽

Data Sets ◽

Holistic View ◽

Content Type ◽

Public Areas ◽

Collection Methods

Purpose This study aims to strengthen implications about hotel cleaning outcomes by comparing guests’ perception of the amount of contact they have with cleanliness of hotel surfaces. Design/methodology/approach This study used two data-collection methods, a survey and an adenosine triphosphate (ATP) test. Data were collected from recent hotel guests using Amazon Mechanical Turk. Guests were asked to identify hotel surfaces that they touch most frequently. Actual hotel cleanliness was measured using empirical data collected with ATP meters. The two data sets were used to compare guests’ perceptions about the amount of contact they have with actual cleanliness measurements of those hotel surfaces. Findings This study found that amount of guest contact was related to cleanliness of surfaces in guestrooms. Significant differences were found in guest perception between high- and low-touch areas and between guestrooms and hotel public areas. More high-touch areas and higher ATP readings were found in guestrooms than in hotel public areas. Originality/value To the best of the authors’ knowledge this study is the first to compare guest contact with hotel surfaces to a scientific measure of hotel cleanliness. In addition, this study is unique because it assesses guest contact and cleanliness of public areas to provide a holistic view of hotel-cleaning needs. The study offers industry empirically based results from guest perception and scientifically based data that can be used to improve hotel housekeeping programs.

Download Full-text

Jointly Improving Parsing and Perception for Natural Language Commands through Human-Robot Dialog

Journal of Artificial Intelligence Research ◽

10.1613/jair.1.11485 ◽

2020 ◽

Vol 67 ◽

pp. 327-374 ◽

Cited By ~ 2

Author(s):

Jesse Thomason ◽

Aishwarya Padmakumar ◽

Jivko Sinapov ◽

Nick Walker ◽

Yuqian Jiang ◽

...

Keyword(s):

Active Learning ◽

Natural Language ◽

Mobile Robot ◽

Real World ◽

Training Data ◽

Mechanical Turk ◽

Amazon Mechanical Turk ◽

Robotic Platform ◽

Language Understanding ◽

Robotic Sensors

In this work, we present methods for using human-robot dialog to improve language understanding for a mobile robot agent. The agent parses natural language to underlying semantic meanings and uses robotic sensors to create multi-modal models of perceptual concepts like red and heavy. The agent can be used for showing navigation routes, delivering objects to people, and relocating objects from one location to another. We use dialog clari_cation questions both to understand commands and to generate additional parsing training data. The agent employs opportunistic active learning to select questions about how words relate to objects, improving its understanding of perceptual concepts. We evaluated this agent on Amazon Mechanical Turk. After training on data induced from conversations, the agent reduced the number of dialog questions it asked while receiving higher usability ratings. Additionally, we demonstrated the agent on a robotic platform, where it learned new perceptual concepts on the y while completing a real-world task.

Download Full-text