TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Amazon’s Mechanical Turk Has Reinvented Research

178 点作者 DmenshunlAnlsis大约 7 年前

12 条评论

Radim大约 7 年前
Mechanical Turk is great for &quot;open&quot;, public research. We used to use them a lot for machine learning tasks (data cleanup, model comparisons, label annotations), along with similar services like CrowdFlower &#x2F; Figure Eight. We saw two primarily issues when applied to &quot;non-open&quot; (commercial) projects:<p>- business-related data too sensitive to share with strangers (contractual obligations, too much risk)<p>- some tasks required non-trivial subject matter expertise and context to annotate properly (quality control issues)<p>For this reason, we gradually moved to an in-house team of long-term annotators. It&#x27;s not much more expensive (moms on maternity leave, students…), but infinitely more flexible and safer for our purposes. YMMV.
评论 #17100218 未加载
评论 #17101900 未加载
评论 #17111581 未加载
评论 #17103570 未加载
评论 #17101345 未加载
评论 #17100121 未加载
评论 #17105935 未加载
rococode大约 7 年前
MTurk is a godsend for ML research and is a huge game-changer. For every other project where the problem is &quot;that sounds cool but we don&#x27;t have enough labeled data&quot; the answer nowadays is &quot;just turk it&quot;. Sentiment labeling, qualitative comparison, error identification, and tons of other traditionally data-scarce tasks are made trivially easy (at the cost of some money) with MTurk, and it&#x27;s pretty much a win-win for everyone involved too!<p>Now the ethics as far as exploitation are definitely important, but I think the design of the site handles things quite well and makes everything fair for all parties. If you feel a task is underpaid, there are enough alternatives that you can just not do it. It&#x27;s also true that there are many international turkers for whom $8&#x2F;hour or less is still solid pay. Then there are also many third-party tools that allow turkers to see which HIT (task) requesters have good track records (low rejection ratio, good pay, etc.), and the site&#x27;s own tools allow requesters to avoid turkers with bad track records. In my experience just browsing through tasks, heavily underpaid tasks don&#x27;t tend to get done (for example, writing a 100-word summary for $0.50).
评论 #17103588 未加载
评论 #17099403 未加载
评论 #17100608 未加载
chx大约 7 年前
Data cleanup is also a great use. For example, we needed to parse an amount of address data, we just needed country, state (or equivalent), city, we gave out each address three times, whatever results were the same at least twice was accepted. We had over 92% where all three were the same, another 7% with 2-1 (required review), less than 1% needed either manual cleanup before re-Turking or just manually entering some of the more gnarly cases. We considered it a truly massive success, price efficient and absolutely unbelievable quick.
评论 #17099472 未加载
评论 #17099660 未加载
ehead大约 7 年前
It&#x27;s an interesting question, whether or not the data is representative (or what demographic is the data representative of).<p>Are there any studies done of the demographic distribution of Turkers?
评论 #17099177 未加载
评论 #17099168 未加载
jankotek大约 7 年前
Please fix misleading title, it should be: Reinvented _Survey_ Research.<p>&gt; <i>thirty to forty minute survey ... paid $1.10</i><p>What sort of results do you expect from this &quot;research&quot;? Do you really expect people to read the questions and answer truthfully?<p>You can prove pretty much anything, by reordering fields or manipulating the questions.
评论 #17100367 未加载
评论 #17099437 未加载
myro大约 7 年前
I&#x27;ve been turking once, made a $100 worth of Amazon credits and bought a first version of Kindle :) good old days
评论 #17101041 未加载
jonbarker大约 7 年前
To paraphrase The Simpsons: &quot;In the future, computer programs will be built by labeled data sets. And our job will be to build and maintain those labeled data sets.&quot;
Nasrudith大约 7 年前
I can&#x27;t help but imagine unscrupulous bot programmers going through every possible survey and answering them quickly with garbage. Or if they monitor expected times doing parallel instances while waiting long enough to look like a human. High volume low value junk ruining things has a long history with the internet.
评论 #17102687 未加载
knguyen0105大约 7 年前
Have anyone used Mechanical Turk for tasks that involve non-English languages like Chinese or Vietnamese?
评论 #17100846 未加载
ada1981大约 7 年前
My friend has used this extensively for all kinds of interesting things.<p>I interviewed him about his work combining mturk &amp; AI to help Trump &#x2F; Clinton supporters better understand each other.<p><a href="https:&#x2F;&#x2F;huffpost.com&#x2F;us&#x2F;entry&#x2F;us_581a4825e4b0f1c7d77c9555" rel="nofollow">https:&#x2F;&#x2F;huffpost.com&#x2F;us&#x2F;entry&#x2F;us_581a4825e4b0f1c7d77c9555</a><p>With regard to pay, it seems reasonable to adopt a standard where the average per hour rate is disclosed in the research papers. This alone may provide social pressure for academics to adopt payment inline with local norms.<p>Also this brings up some questions for me. Is it unethical for a researcher in a locale with low wages to post on mturk looking for work at comparable rates? Should posters be posting rates comparable to their own countries minimum wage laws? Is there another standard? Could requiring some researchers (from richer countries) to increase their rates result in resesrchers from other countries being priced out or having their research deprioritized by Turks?
8bitsrule大约 7 年前
&#x27;Mechanical Turk&#x27; is a sweatship run by a notorious skinflint.
ccnafr大约 7 年前
Do people who participate in MK get any money for their time?
评论 #17100508 未加载
评论 #17100826 未加载