Kathy Reid<p>Delighted to be able to publicise a paper that was presented at the @ALTAnlp 2023 Workshop at the end of last year, co-authored with my <a href="https://aus.social/tags/PhD" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>PhD</span></a> supervisor, Associate Professor @eltwilliams, and written as part of my research at <a href="https://aus.social/tags/ANU" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ANU</span></a> School of Cybernetics. </p><p>Titled "Right the docs: Characterising voice dataset documentation practices used in machine learning", it combines both exploratory interviews and documentation analysis to characterise how large voice datasets - e.g. <a href="https://aus.social/tags/LibriSpeech" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LibriSpeech</span></a>, <span class="h-card" translate="no"><a href="https://mozilla.social/@mozilla" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>mozilla</span></a></span>'s <a href="https://aus.social/tags/CommonVoice" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>CommonVoice</span></a>, and several others, document their <a href="https://aus.social/tags/metadata" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>metadata</span></a>. </p><p>Unsurprisingly, it finds that the <a href="https://aus.social/tags/dataset" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>dataset</span></a> <a href="https://aus.social/tags/documentation" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>documentation</span></a> practices seen currently do not meet the needs of the <a href="https://aus.social/tags/ML" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ML</span></a> practitioners who use these datasets.</p><p>We show, once again, in the words of Nithya Sambasivan - "everyone wants to do the model work, but nobody wants to do the data work" ...</p><p><a href="https://aclanthology.org/2023.alta-1.6/" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">aclanthology.org/2023.alta-1.6</span><span class="invisible">/</span></a></p><p><a href="https://aus.social/tags/RightTheDocs" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>RightTheDocs</span></a> <a href="https://aus.social/tags/WriteTheDocs" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>WriteTheDocs</span></a></p><p>Citation: </p><p>Reid, K., Williams, E.T., 2023. Right the docs: Characterising voice dataset documentation practices used in machine learning, in: Muresan, S., Chen, V., Casey, K., David, V., Nina, D., Koji, I., Erik, E., Stefan, U. (Eds.), Proceedings of the 21st Annual Workshop of the Australasian Language Technology Association. Association for Computational Linguistics, Melbourne, Australia, pp. 51–66.</p>