Menu
** Advanced Search **
I’ve been working on implementing an advanced search feature. You can test it here: https://dev.tatoeba.org/sentences/search
Now, unapproved and orphan sentences are no longer put at the end of the search results. Instead, they are filtered out by default when performing a search from the top bar. One can later make them appear in the results by changing the appropriate criterion.
Feel free to comment on anything. Note that searches are performed on a copy of tatoeba.org’s database from February 2nd, so you won’t find sentence added past this date. But you can add new sentences just for the sake of searching. They should be visible within ten minutes.
I’d like you to test not only all the criteria, but also searching for newly-added or newly-modified sentences. I changed the way new or modified sentences got indexed, so this part may contains bugs too. In short, every sentence you add or modify should become visible (or disappear) in a search within ten minutes, be it because of a modification of its contents, its tags, ownership, audio, link etc. anything that can be searched.
Previous thread: https://tatoeba.org/wall/show_message/22852
[not needed anymore- removed by CK]
I tried to put the search fields to the right, although it feels a bit packed to me, and it requires quite some scrolling to get to the submit button, and it uglify the page when there are no results. Tell me what you think.
I prefer the original layout, for the reasons you say. (Note that it requires scrolling not only to hit the submit button, but also to access many fields that I would expect would be frequently changed.) However, if you do put it on the right, I would hope that you would at least keep the link to an advanced search, since, as it is now, there's no clue that we support all these additional criteria.
[not needed anymore- removed by CK]
[not needed anymore- removed by CK]
Yes. Also, would it be possible to open up a new window (or tab -- I guess this would depend on how a user's browser is configured) for search results? I've wanted this before, but with the advanced search feature, it makes even more sense. Then one could keep open the search criteria window/tab for future searches even after one has closed the the results window/tab. In that case, the layout that would require less scrolling would be the original "horizontal" one rather than the right-sidebar "vertical" one.
There's a typo: "Oprhan sentences are likely to be incorrect."
Where?
When you type right-to-left text into the "Words" field, it's left-justified rather than right-justified. This differs from the behavior of, say, the "Example sentences with the words:" field in the main search area.
It would be nice to have a way to "gray out" the whole "Translations" box for those times when you don't want it to affect your search.
It would also be nice to have a "Randomize" feature that displays the sentences in random order so that you don't end up seeing the same sentences over and over again, especially if it produces a search that you want to bookmark and use repeatedly. I did this in a bookmarklet that I wrote ( http://en.wiki.tatoeba.org/arti...ated-sentences ).
In general, the new feature works very well.
What do you mean by graying out? Do you mean a reset button for that part only? Or do you mean to gray it out automatically when its fields are all set their default values?
I guess a reset button is what I want. Better not to gray it out, since that might make people think that nothing in that area could be selected.
[not needed anymore- removed by CK]
Since adding a “random page” link on every pagination is rather more difficult to do, I only added a random sort order function to the advanced search.
It’s now possible to randomize the results by selecting the appropriate option of the “Sort” field.
I love it, including the other sorting options. Thanks!
I suggest these rewordings:
"Results sort" -> "Order" (because the box caption already says "Sort")
"Less number of words should be first" -> "Shortest first" or "Fewest words first"
"Last created first" -> "Most recently created first" (if it fits, otherwise keep what you have)
"Last modified first" -> "Most recently modified first" (if it fits, otherwise keep what you have)
Also, a rewording for the "More search criteria" group box:
"Owned by a self-proclamed native" -> "Owned by a self-identified native" or "Owned by a self-identified native speaker"
And don't forget this one:
"Owned by a self proclaimed hyper-genius allegedly being fluent at a near-native level in fifty languages"
;-)
[not needed anymore- removed by CK]
This is strange indeed.
https://dev.tatoeba.org/jpn/sen...io=&sort=words
I don't mind seeing sentences owned by non-native speakers though, as long as they are tagged OK by native speakers. I'd find the criterion "owned or tagged OK by self-proclaimed native speakers" more useful.
[not needed anymore- removed by CK]
You're right, but I still think that the mother tongue of a sentence has nothing to do with its quality. Are you implying that sentences that are in your OK list and owned by non-native speakers are worse in quality than those owned by native speakers?
It'd be also nice if we could search for/among the sentences owned or tagged OK by one or more particular members.
That’s because dev.tatoeba.org’s database is so old that doesn’t contain much natives. There are only people who set themselves as natives on dev.tatoeba.org when Trang introduced this feature. See https://dev.tatoeba.org/stats/native_speakers
[not needed anymore- removed by CK]
> 1. Another idea to consider would be to allow the option of an alphabetical sort if it doesn't put too much of a load on the server.
> 2. Also, another idea to consider would be to allow the option to sort the sentences by how they end, if this doesn't put too much load on the server.
I think this would be too resource-hungry as we’d need to index all the sentence texts.
> 3. This one may put too much load on the server, but ...
This won’t put any load and is easy to do.
[not needed anymore- removed by CK]
No, it’s not possible to display more than the first 1000 results. Why would you want that by the way? Also, what if the search criteria are set so that sentences disappear from the search results once translated? This way, the contributor only needs to translate the first results.