Jump to content

Talk:Crowdsourcing

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia


Wiki Education Foundation-supported course assignment

[edit]

This article is or was the subject of a Wiki Education Foundation-supported course assignment. Further details are available on the course page. Peer reviewers: Emmariell.

Above undated message substituted from Template:Dashboard.wikiedu.org assignment by PrimeBOT (talk) 18:44, 16 January 2022 (UTC)[reply]

Collaborative Mapping

[edit]

Added collaborative mapping in "see also". Do not know, if a seperate small section for "collaborative mapping" will be appropriate here. Mapping products like Open Street Map is not covered explicitely as crowdsourcing objective (like Humanitarian Openstreetmap Team) and may be extending the collaborative mapping article is the better place.--Bert Niehaus (talk) 09:21, 6 October 2018 (UTC)[reply]

Crowdsourcing in machine learning

[edit]

Hello everyone,

We would like to suggest changes to this article. The crowdsourcing approach is successfully applied in machine learning as well. Researchers and practitioners in machine learning use crowdsourcing techniques to collect training data sets for ML models of various data science applications. In contrast to traditional in-house editorial-based data collection methods that are usually slow and expensive, crowdsourcing helps them to get data sets in a very short period of time and in a fairly inexpensive way.

In particular, to train self-driving cars to recognize objects on roads, engineers collect labeled images with people, road signs, traffic lights, and other objects on the streets. Voice assistants are trained with audio recorded and labeled by thousands of crowd workers with different voices and accents. For developing a search engine ranking model, engineers ask crowd workers to assess if search engine results match user queries. Among other applications that are trained on labeled data sets are translators, optical character recognition programs, chatbots, navigators, taxi mobile apps, etc.

Data labeling tasks can be designed and programmed into an open crowdsourcing platform. These platforms provide on-demand access to a large crowd of workers and some tools that help to control the quality of responses. Another common way is to use internal tools, such as internal crowdsourcing platforms, or ready-made solutions.

References:

1. Alonso O. (2019). The Practice of Crowdsourcing. Ed. by G. Marchionini. Synthesis Lectures on Information Concepts, Retrieval, and Services. Morgan & Claypool Publishers. https://www.google.ru/books/edition/The_Practice_of_Crowdsourcing/rq2aDwAAQBAJ?hl=ru&gbpv=0&bsq=Omar%20Alonso

2. Alexey Drutsa, Viktoriya Farafonova, Valentina Fedorova, Olga Megorskaya, Evfrosiniya Zerminova, and Olga Zhilinskaya. 2019. Practice of Efficient Data Collection via Crowdsourcing at Large-Scale. Tutorial at the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (KDD '19). https://arxiv.org/pdf/1912.04444.pdf

3. Nikita Pavlichenko, Ivan Stelmakh, Dmitry Ustalov. 2020. CrowdSpeech and Vox DIY: Benchmark Dataset for Crowdsourced Audio Transcription. NeurIPS 2021 Datasets and Benchmarks Track (Round 1). https://openreview.net/forum?id=3_hgF1NAXU7 — Preceding unsigned comment added by Crowdfollowsme (talkcontribs) 11:38, 6 October 2021 (UTC)[reply]

--Crowdfollowsme (talk) 12:01, 6 October 2021 (UTC)[reply]

Well, there were multiple bold attempts to add such section recently, e.g. https://en.wikipedia.org/w/index.php?title=Crowdsourcing&oldid=1038076350, which were reverted as they looked as a very opionated claims regarding ML (e.g. the crowdsouring approach being applied by the majority of products based on ML), written as a reason to cite Yandex/Toloko platform, to which you are affiliated, and some selected online courses from Internet. At the same time there is an existing sub-section "Microwork" which I believe describe the same stuff, citing Amazon's MTurk. I'd say, you better write about Yandex/Toloko in there, expanding the section, but avoiding claims that this is the only/best way to do ML. Birdofpreyru (talk) 20:21, 6 October 2021 (UTC)[reply]

Length

[edit]

@BeenAroundAWhile: Greetings! I see you tagged this page for length a few years ago. The page statistics show 47k of readable prose, which is on the short end according to WP:SIZERULE. I quickly skimmed it and didn't see anything obviously overflowing...were there any parts in particular that you think could be moved to a detail article? Feel free to ping me and I can look into it. I'd like to resolve the length question soon so the article can be copyedited. -- Beland (talk) 00:34, 26 January 2022 (UTC)[reply]

Wiki Education assignment: Research Process and Methodology - SU22 - Sect 202 - Tue

[edit]

This article was the subject of a Wiki Education Foundation-supported course assignment, between 4 July 2022 and 16 August 2022. Further details are available on the course page. Student editor(s): WengConor (article contribs).

— Assignment last updated by Sss88891 (talk) 00:26, 27 July 2022 (UTC)[reply]

Wikipedia Ambassador Program course assignment

[edit]

This article is the subject of an educational assignment at Carnegie Mellon supported by Wikipedia and the Wikipedia Ambassador Program during the 2012 Spring term. Further details are available on the course page.

The above message was substituted from {{WAP assignment}} by PrimeBOT (talk) on 16:00, 2 January 2023 (UTC)[reply]