Search

This paper is interested in investigating whether human gaze signals can be leveraged to improve state-of-the-art search engine performance and how to incorporate this new input signal marked by human attention into existing neural retrieval models. In this paper, we propose GazBy ({\bf Gaz}e-based {\bf B}ert model for document relevanc{\bf y}), a light-weight joint model that integrates human gaze fixation estimation into transformer models to predict document relevance, incorporating more nuanced information about cognitive processing into information retrieval (IR). We evaluate our model on the Text Retrieval Conference (TREC) Deep Learning (DL) 2019 and 2020 Tracks. Our experiments show encouraging results and illustrate the effective and ineffective entry points for using human gaze to help with transformer-based neural retrievers. With the rise of virtual reality (VR) and augmented reality (AR), human gaze data will become more available. We hope this work serves as a first step exploring using gaze signals in modern neural search engines.

Information about the uncertainty associated with Earth science observational data is fundamental to use, re-use, and overall evaluation of the data being used to produce science and support decision making. The associated uncertainty information leads to a quantifiable level of confidence in both the data and the science informing decisions produced using the data. The current breadth and cross-domain depth of understanding and application of uncertainty information, however, are still evolving as the practices associated with quantifying and characterizing uncertainty across various types of Earth observation data are diverse. Since its re-establishment in 2015, the Information Quality Cluster (IQC) of the Earth Science Information Partners (ESIP) has convened numerous sessions within the auspices of ESIP and the American Geophysical Union (AGU) to help collect expert-level information focusing on key aspects of uncertainty of Earth science data and addressed key concerns such as: 1) how uncertainty is quantified (UQ) and characterized (UC), 2) understanding the strengths and limitations of common techniques used in producing and evaluating uncertainty information, 3) implications using uncertainty information as a quality indicator 4) impacts of uncertainty on data fusion/assimilation, 5) various methods for documenting and conveying the uncertainty information to data users, and 6) understanding why certain user communities care about uncertainty and others do not. A key recommendation and action item from the ESIP Summer Meeting 2017 was for the IQC to develop a white paper to establish a clearer understanding of the concept of uncertainty and its communication to data users. The information gathered for this white paper has been provided by Earth science data and informatics experts spanning diverse disciplines and observation systems in the cross-domain Earth sciences. The intention of this white paper is to provide a diversely sampled exposition of both prolific and unique policies and practices, applicable in an international context of diverse policies and working groups, made toward quantifying, characterizing, communicating and making use of uncertainty information throughout the diverse, cross-disciplinary Earth science data landscape.

The article of record as published may be found at https://doi.org/10.7930/