Chrome Extension
WeChat Mini Program
Use on ChatGLM

Astro Data Lab Spectral Viewer Requirements for Wide-Area Spectroscopic Surveys

arxiv(2023)

Cited 0|Views45
Abstract
The Astro Data Lab is preparing to host large spectroscopic datasets such as a copy of the Dark Energy Spectroscopic Instrument (DESI) survey, which is projected to include approximately 40 million spectra of galaxies and quasars as well as over 10 million spectra of stars by 2026. Currently, we serve DR16 spectra from the Sloan Digital Sky Survey (SDSS), including Baryon Oscillation Spectroscopic Survey (BOSS), and Extended BOSS (eBOSS) spectra. A spectral viewer tool allows users to visually and interactively inspect spectra. Given the large size of these spectroscopic datasets, a typical use case might consist of a selection or query for a subset of objects of interest (e.g., a subsample of stars or galaxies or quasars), followed by visual inspection of the selected spectra. It is anticipated that in some cases, users will want to go through a long list of spectra (e.g., thousands) quickly while looking for specific features. This document contains a description of the requirements for such a spectral viewer tool to be incorporated within the Astro Data Lab environment at NSF's NOIRLab. For each object, the spectral viewer will display the observed spectrum and, if available, the noise spectrum, sky spectrum, and best-fit template spectrum. Users will be able to control the display interactively after they launch the tool as part of their Data Lab workflow. The primary objective will be to support the visualization of spectroscopic datasets hosted at the Astro Data Lab but this requirements document could be a useful reference or inspiration for other applications and/or other datasets in the astronomy community.
More
Translated text
PDF
Bibtex
AI Read Science
Must-Reading Tree
Example
Generate MRT to find the research sequence of this paper
Data Disclaimer
The page data are from open Internet sources, cooperative publishers and automatic analysis results through AI technology. We do not make any commitments and guarantees for the validity, accuracy, correctness, reliability, completeness and timeliness of the page data. If you have any questions, please contact us by email: report@aminer.cn
Chat Paper

要点】:本文提出SD-Eval,一个针对spoken dialogue理解与生成的多维度评估基准数据集,强调对副语言和环境信息的考量,通过实验验证了加入这些信息的模型性能提升。

方法】:通过整合八个公共数据集,构建了一个包含7,303个发言、8.76小时语音数据的SD-Eval数据集,涵盖了情感、口音、年龄和背景音四种维度。

实验】:使用三个不同的模型对SD-Eval进行评估,构建的训练集包含1,052.72小时语音数据和724.4k发言,通过客观指标(如BLEU和ROUGE)、主观评价及基于LLM的指标进行综合评价,实验结果显示加入副语言和环境信息的模型在客观和主观测量上均优于对照组,且LLM-based指标与人类评价的相关性更高。数据集开源地址为https://github.com/amphionspace/SD-Eval。