Comments on: Using internet users to do something useful http://www.mblondel.org/journal/2009/06/13/using-internet-users-to-do-something-useful/ Computer science, Chinese, Japanese, random thoughts… Thu, 09 Sep 2010 19:22:33 +0000 http://wordpress.org/?v=2.5.1 By: How millions of internet surfers, including you, unknowingly help digitize ancient books | Hermes Technologies Ltd. http://www.mblondel.org/journal/2009/06/13/using-internet-users-to-do-something-useful/#comment-196648 How millions of internet surfers, including you, unknowingly help digitize ancient books | Hermes Technologies Ltd. Tue, 15 Dec 2009 12:22:37 +0000 http://www.mblondel.org/journal/?p=109#comment-196648 [...] Blondel explains the idea behind a CAPTCHA in his article on using internet users to do something useful: The idea of CAPTCHA is to ask users to answer a question that only humans can answer in order to [...] [...] Blondel explains the idea behind a CAPTCHA in his article on using internet users to do something useful: The idea of CAPTCHA is to ask users to answer a question that only humans can answer in order to [...]

]]> By: Mathieu http://www.mblondel.org/journal/2009/06/13/using-internet-users-to-do-something-useful/#comment-187093 Mathieu Tue, 16 Jun 2009 07:48:50 +0000 http://www.mblondel.org/journal/?p=109#comment-187093 Interestingly enough, LeMonde.fr had an article about this today... http://www.lemonde.fr/technologies/article/2009/06/10/quand-les-joueurs-de-jeux-video-instruisent-les-intelligences-artificielles_1201417_651865.html#ens_id=1201461 Interestingly enough, LeMonde.fr had an article about this today…

http://www.lemonde.fr/technologies/article/2009/06/10/quand-les-joueurs-de-jeux-video-instruisent-les-intelligences-artificielles_1201417_651865.html#ens_id=1201461

]]> By: Mathieu http://www.mblondel.org/journal/2009/06/13/using-internet-users-to-do-something-useful/#comment-186951 Mathieu Sat, 13 Jun 2009 14:29:50 +0000 http://www.mblondel.org/journal/?p=109#comment-186951 Yes some people will have to mark whether the input characters are correct or not. Either this can be done by "proofreaders" or we can make this process part of the game itself. You mention that it may be faster for the proofreaders to rewrite the character by hand but we need our sample database to include various writing styles so it's better to collect samples from as many people as possible. Also the user interface can be designed to make the verification process as smooth as possible (stroke order decomposition, keyboard shortcuts, ...) We can also use an existing recognizer like Zinnia and verify in priority those which were not correctly recognized... Yes some people will have to mark whether the input characters are correct or not. Either this can be done by “proofreaders” or we can make this process part of the game itself. You mention that it may be faster for the proofreaders to rewrite the character by hand but we need our sample database to include various writing styles so it’s better to collect samples from as many people as possible. Also the user interface can be designed to make the verification process as smooth as possible (stroke order decomposition, keyboard shortcuts, …) We can also use an existing recognizer like Zinnia and verify in priority those which were not correctly recognized…

]]> By: Christoph http://www.mblondel.org/journal/2009/06/13/using-internet-users-to-do-something-useful/#comment-186938 Christoph Sat, 13 Jun 2009 08:30:38 +0000 http://www.mblondel.org/journal/?p=109#comment-186938 Handwriting recognition can be helpful for learners to train new characters and have them evaluated by the recognition engine. Now, would this game fit into the category of "games with a purpose"? We want to judge the user's input with unsufficient data, and we want to improve our insufficient data with uncertain input. I guess manually labeling user input as good or bad is not a way out of this dilemma, as the process of verification will take longer than redrawing the character by hand. Yust my random thought... Handwriting recognition can be helpful for learners to train new characters and have them evaluated by the recognition engine. Now, would this game fit into the category of “games with a purpose”? We want to judge the user’s input with unsufficient data, and we want to improve our insufficient data with uncertain input.
I guess manually labeling user input as good or bad is not a way out of this dilemma, as the process of verification will take longer than redrawing the character by hand.
Yust my random thought…

]]>