Adostrans contributed definitions quality issues?

[msittig]

Is it just me, or has there been an increase in bad definition being added to the Adsotrans database recently? I keep getting ridiculous ones that are clearly context-dependent, ie not appropriate for a general context.

For example, I just got "decided" for 斩断, which literally means to cut off or behead. I could see it meaning "decided" in some figurative context, but that conversion from literal meaning -> figurative meaning should be made in the mind of the reader, not in the Adso database.

Just felt like ranting a bit. It's this is a non-issue for everyone else, please ignore me.

[muyongshi]

I don't use Adsotrans for this reason. I find too many words are not accurately translated or some sap just put what he THOUGHT it should be....

[Lu]

I rarely ever use it, for the same reason. It seems to know less than I do, so it's not a very useful dictionary for me. It once translated 西 as 'Spanish', speaking of context-related.

It might get better over time though.

[msittig]

我是

PART OF SPEECH = PERSON

ENGLISH = I am

I mean seriously.

Edit → Search → Delete.

[zhwj]

Isn't there a periodic manual review and commit process that would eliminate these problems? I know that changes are put into place live for the working database, but I seem to recall that there's another level of screening before they're made permanent.

I guess that's just the nature of crowd-sourcing - you either have immediate results, trusting that people will use it responsibly but running the risk of bad data, or you hold everything and moderate it before releasing it, ensuring quality but making people wait to see the results of their efforts in the project.

[roddy]

I don't use Adsotrans as I find that if something isn't in my dictionaries a smart Google search is my best bet. Just had a look at a paragraph of a document which I'll be translating tomorrow though, and I thought it did pretty well. The only two issues were that it's currently translating 的 as di3, base (?), and 曾让 was picked up as a name.

[msittig]

Nowadays I use Google Translate to do a rough translation, and Adso (along with other dictionaries, Google searches, Wenlin) only as a secondary backup.

Adso's user-contributions are its greatest strength and weakness.

[tooironic]

I find this annotator http://www.chinese-tools.com/tools/annotation.html much better than adsotrans myself.

[trevelyan]

Micah,

We're seeing a lot more editing taking place. This means more contributions both good and bad. All edits are still logged for review. I'm behind schedule on updating the master database though, and don't want to update it until I've caught up since I don't want people to find their contributions have "disappeared" overnight.

I'm open to suggestions on how to deal with this. I don't want to make it more difficult to contribute since immediate feedback matters and I don't want to discourage contributions. If quality is becoming an issue in the meantime, perhaps the solution is setting up a "stable" interface that uses a non-live copy of the database?

[trevelyan]

@muyongshi, @tooironic -- if you guys have better sources of data please let me know and I'll incorporate them. I'm not familiar with anything that can be legally shared that is more comprehensive or accurate than what we have.

It's a legitimate complaint if a slow editing process is degrading the quality of the online engine. I'm not sure what the proper solution is -- our goal is better semantic analysis and the creation of resources that people need. If being open helps us get there faster than being closed that's the route we're going to take.

[trevelyan]

I've just put up a page that annotates text using the last stable database release. This redirects the request to a clean database rather than the one currently being edited. Editing functions still work, although the edits are applied to the current database.

http://adsotrans.com/stable.html

Added a link to this version on the main page as well.