« first day (63 days earlier)      last day (2689 days later) » 

3:47 AM
@MartinSleziak Hi Martin, just an update, I checked the indexed content of post 985338 (docID=483608 in my index), its title is "Remarks on a Previous Post", and its content is merely one sentence "So, is there any elementary method to prove this?"
According to the revisions page, I think it is my crawler or indexer missed part of its content. I do not know why, but it is not searcher's bug, which means the problem can be easier to locate. I am going out now and will investigate this further.
 
Thanks for the update.
This is unrelated to this issue but I have question about AND vs. OR:
11 hours ago, by Wei Zhong
@MartinSleziak I assume this is not a bug. First, by "adding something more to the query" you mean adding another keyword, by default, Approach0 uses OR instead of AND to search keywords. If you get that post by using the second query, then the first keyword ($n>\dfrac{p_n}{\ln p_n}$) perhaps does not contribute to your search results at all. (This is also indicated by the search results without any math expression being highlighted)
So if I understand this correctly, Approach0 uses OR by default, if I enter several formulas or several phrases (or combination of both).
Is there a way how I can tell in the search query that I want to use AND? Or am I only able to use OR in this search engine?
 
 
6 hours later…
9:39 AM
@MartinSleziak, Approach0 looks awesome. Thank you for letting me know such a nice search engine! — Sangchul Lee 29 secs ago
@SangchulLee You can find links to some further information here. — Martin Sleziak 8 secs ago
 
9:53 AM
@WeiZhong One could just start a Math in Chinese proposal at Area51, the same way that Stackoverflow was proposed in other languages, like here. If it receives enough attention, then it may definitely be a success, and it wouldn't require you to make a Q&A site from scratch.
@MartinSleziak Here's my template: Duplicate/Related: [$$](). (*Found using [Approach0.xyz](https://approach0.xyz/)*)
 
@Workaholic By template you mean template to copy and paste right? Or do you have some kind of bookmarklet which generates this including links?
I sometimes use bookmarklet which generates links like this for the page I am viewing by one click: [Announcing a third-party search engine for Math StackExchange.](http://meta.math.stackexchange.com/q/24978) Announcing a third-party search engine for Math StackExchange.
Unfortunately, it gest bungled up when the title contains MahJax. :-(
So in such cases I am often too lazy to edit it (as in the comment above.)
 
 
2 hours later…
11:58 AM
@MartinSleziak Yes just copy/paste. Unfortunately, it's all manual, it would be very nice if there was a way to automate such process.
 
Thanks for the reply.
 
Just noticed it: Approach0 treats \rightarrow the same way as \to, very nice! I had to individually change them each time when using traditional search engines to squeeze some result.
 
@Workaholic Unfortunately I never programmed something like that. One math.SE user made many useful bookmarklets. The site with them was in the meantime deleted, some of them might be probably retrieved from here.
Personally I am only using Comments bookmarklet (copies all comments, including MarkDown). And Link bookmarklet (but for some time it does not work well if the title has MathJax - I am not sure whether it works for other users).
In fact, I only found that archive thanks to this: Save the answer in TeX/PDF - revisioned
 
@MartinSleziak I don't know how to code Javascript but maybe I can tweak one of those scripts, let me try...
 
Well I am not sure that it would be that easy.
It might be worth trying. But personally, I have plenty of other things to do. :-(
So if I try to modify some of those bookmarklets or create my own, it won't be anytime soon.
 
12:38 PM
javascript:(function()%7Bs%3Ddocument.querySelector('.search-container%20input'‌​).value%3Bu%3DencodeURI('%2F%2Fapproach0.xyz%2Fsearch%3Fq%3D'%2Bs%2B'%20%3A'%2Bdo‌​cument.title).replace(%2F%5C%2B%2Fg%2C'%252B')%3Bwindow.open(u)%7D)()
This is what I was able to get for the moment, it still needs some more work so that only what's inside the $ would be picked for the search..
 
What does it do? Should it be used when I have Approach0 window open? Or when I am in the tab with question?
 
@MartinSleziak Automatically takes the window's title and searches it in approach0
 
12:50 PM
I was not able to make it work. But something happens (for a very short moment something appears in the address bar).
I tried to allow pop ups. I will probably play with that later, when I have more time.
 
You were in a tab with math.SE correct?
 
But thanks a lot for you effort on creating new useful bookmarklet.
Yes, I was.
@Workaholic It works now.
I feel really stupid.
I was trying whether I should enable popups or something like that.
 
So what was the problem? :P
 
And the only problem was that I copied the source you pasted here with the exception of one character...
Thanks for the bookmarklet once again.
 
You're welcome! It's not that useful in its current state but I will do my best to try to ameliorate it.
 
12:55 PM
It is still probably faster to make one click, then click on raw query and edit it there.
Depends on the question, but if it has relevant equation(s) in the title, it's faster than retyping it.
 
Indeed.
 
 
2 hours later…
3:15 PM
@MartinSleziak You usually do not need to specify AND operator, because those match both (or most) keywords will automatically go to the top of the result list, and those match less number of query keywords will most likely be ranked lower. In fact, Approach0 does not support explicit AND operator currently.
 
Ok, so if I got it right, the top results will be similar to AND and then eventually I get to results which are closer to OR?
Very roughly speaking.
 
@Workaholic Some of you probably do not know that StackExchange is not easily accessible in China because it is using JQuery (a Javascript library) hosted on googleapis.com which is blocked in China.
@MartinSleziak Correct, if you see all the top results only match one or a few keywords, most likely (although in some rare cases when term-frequency is really large, the result is heavily weighted by term-frequency instead of by the variety of matching terms) there is no result that matches ALL the terms.
 
Ok. Thanks for the reply!
 
@WeiZhong Oh, yes, forgot that. I saw a thread somewhere in meta.SE that mentioned that. No idea why the SE team would need a 3rd party instead of immediately hosting it.
 
@Workaholic Yes, if SE team can host all these 3rd party libraries (very small, not a big deal) on SE itself, then it will benefit a lot of users from China.
I wonder if there is a way to at least let SE team know about this issue. Maybe someone would help me draft a post to sign a request on SE meta.
 
3:33 PM
@WeiZhong Found the thread I was talking about, meta.stackoverflow.com/questions/258288/…
 
@Workaholic First time to know that you are actually capable of tweaking some Javascript, nice job!
 
@WeiZhong Haha, just trial and error! :)
 
@Workaholic Me too when I was programming :P
 
@WeiZhong They could also have a separate CDN that hosts that JS library that loads only with Chinese users, so that the load wouldn't be that much, I don't know too much about this but I think that's probably possible.
 
@Workaholic After a quick read of that post, I disagree the technical reasons listed by SE developers too. As the second answer suggests, Javascript library CDN can be hosted by other professional CDN services as well. I think there are a number of services that provide good CDN service across China and inside China. AliYun CDN is one example.
@Workaholic Exactly what I am thinking.
 
3:45 PM
@WeiZhong You could try to contact them concerning that, IMHO SE has everything to win if they make such decision.
 
I assume there is no big commercial benefit from their perspective to drive them and make this change.
 
But they will have a much more users, which is in their benefit.
 
Isn't there a feature request about this on meta SE? (I would expect that, but I did not find it.) I have already upvoted the post linked by Workaholic.
 
@Workaholic Not exactly, few Chinese people know SE, average people are reluctant to read English in the first place. Although a considerable number of programmers in China know SO, but few know SE is their mother company.
 
@MartinSleziak It's meta.stackoverflow, maybe @WeiZhong may make a feature request post on this @ meta.SE
 
3:49 PM
I can tell you that the most popular sites in China about math Q&A is some mobile APPs that help student search homework questions. (to my knowledge, they are using deep learning to process photos uploaded by students) And these APPs are getting extremely popular in China
Some examples are 小猿搜题(yuansouti.com),guangzi Q&A (guangzixuexi.com/wenda), afanti(afanti100.com) and zuoyebang (zuoye.baidu.com)
Not like the atmosphere on SE (MSE in particular), where people asking homework questions will not be encouraged. It is difficult for some of you to accept these APPs I assume, also a headache for many teachers and parents in China AFAIK.
 
Interesting.
 
Yet many people view math.SE as the place which helps students to cheat at homework: Homework and migrating to math.se at meta.MO.
 
@Workaholic Based on the reactions towards the already proposed one on meta, I think I lost my interest to bring this topic up again. At least for now I doubt the open of MSE to China will attract enough Chinese users to a degree that will make commercial sense to SE.
 
@WeiZhong But it was proposed 2 years ago, maybe the SE team may get interested in that now.
 
4:05 PM
@MartinSleziak When talking about commercial interests, helping student do their math homework and help programmers get shit things done is indeed the most attractive place for companies doing K12.
 
I have upvoted at least this feature request on meta.SE which seems related: Add a fallback for jQuery in case it fails to load from Google's CDN (I do not know about the issue to be able to say whether it would solve the problem.)
 
@MartinSleziak Yes, that would solve the problem.
 
Good, but I still think math SE will have little chance to attract users from China if their content data are still mostly in English.
 
@WeiZhong But at least that would solve the problems people get when accessing SE sites in general, including stackoverflow. If things get better you may then try to make a proposal at Area51 for a Chinese version of Math.SE.
 
And I believe a popular Q&A website in China must have operating team and active users from China as well.
@Workaholic OK. What I can do is to join SE meta and also upvote for that post. But I really do not want to create a same feature request again on meta SE. Also it is not meanful to make a proposal at Area51 for Chinese version now if they are still using Javascript CDN from sites blocked by China.
 
4:21 PM
@WeiZhong That sounds like the right plan. :)
 
Sadly I just find they require at least 15 reputation from join a new sub-site to upvote a post. Anyway, I am going to sleep now. Nice to have this conversation, and I hope I will soon get that "bug" fixed.
 
@WeiZhong Yeah, I hope too that they can fix that. Goodbye!
 
4:56 PM
So I'm now approaching something, javascript:(function(){s=document.querySelector("a.question-hyperlink");u=encod‌​eURI('//approach0.xyz/search?q='+s+'+');window.open(u)})() inputs the URL of the question, I will try to find out how to select only the text and not the link itself.
 
5:24 PM
@MartinSleziak Finally a version that works: javascript:(function(){s=document.querySelector("a.question-hyperlink").textCon‌​tent;u=encodeURI('//approach0.xyz/search?q='+s+'+');window.open(u)})()
 
What are the differences from the old version?
 
@MartinSleziak It doesn't take the window title but rather the only question's title, unfortunately it has a problem with the + sign, let me try to fix that
 
5:45 PM
@MartinSleziak Fixed again! :) javascript:(function(){s=document.querySelector("a.question-hyperlink").textCon‌​tent;var%20u=encodeURI('//approach0.xyz/search?q='+s+'&p=1').replace(%2F%5C%2B%2F‌​g%2C'%252B');window.open(u)})()
 

« first day (63 days earlier)      last day (2689 days later) »