2
$\begingroup$

As has been discussed in this meta post, there is a third party search (with $\LaTeX{}$ support) engine for Mathematics Stack Exchange. I was wondering whether there is search engine with similar functionalities for both MathOverflow and Meta-MathOverflow.

My questions are,

  • Are there any such search engines?

  • If not, then can anything be done regarding "installing" such a feature to this website?

$\endgroup$
2
  • 4
    $\begingroup$ Techinically, adding site:mathoverflow.net at the end of a Google search already gives a third-party search engine. If the important point here is the Latex support, you should probably mention it in the question. $\endgroup$ Jan 7, 2017 at 9:35
  • $\begingroup$ @FedericoPoloni: I didn't know that such think can indeed be done. Thanks for the information. $\endgroup$
    – user57432
    Jan 7, 2017 at 17:19

5 Answers 5

7
$\begingroup$

I am the creator of Approach0 (https://approach0.xyz), a math-aware search engine.

Approach0 is currently only indexing Math StackExchange data, however, I am considering to add MathOverflow to Approach0. I think this is a good opportunity for posting a survey here.

Please give Approach0 a try and tell me to which degree do you think it can be helpful to you or MathOverflow users? How often do you find a math-aware search engine desired when you are spending time on MathOverflow?

I also want to see how much interests arising from MathOverflow users for Approach0. I am working on this side-project in spare time, but if enough potential user intersts demonstrated here, I will spend some time to write a crawler to start including MathOverflow data.

$\endgroup$
10
  • $\begingroup$ A naive question: for latex expressions, does it search for the exact character string? I am in the habit of inserting blank spaces in expressions for various reasons, one being readability. But many users don't. $\endgroup$
    – Todd Trimble Mod
    Jan 7, 2017 at 19:05
  • $\begingroup$ @ToddTrimble I suppose Wei Zhong will give you more details but very short answer is no, it is not search for exact string. You might find a bit of explanation in chat, for example here and perhaps also here. $\endgroup$ Jan 7, 2017 at 21:04
  • $\begingroup$ @ToddTrimble Martin is right. $\endgroup$
    – Wei Zhong
    Jan 8, 2017 at 7:55
  • $\begingroup$ @ToddTrimble I would say that a good way to get at least some basic insight into what the capabilities of this search engine are and for what type of searches it is useful, you might tried some queries tested before by other users. You can find such queries for example in comments on math.SE (and to a lesser extent on MO) or in this chatroom. $\endgroup$ Jan 9, 2017 at 8:43
  • $\begingroup$ I tried a few searches: $\int_{-\infty}^\infty e^{-x^2}dx$ was good, $\mathbb{R}P^\infty$ was not bad, $M\otimes N$ gave lots of hits for $MN$ or $m\times n$ which is not so useful, $\Omega S^3$ gave results which made no sense to me at all. Still, the basic framework looks very promising, and I am sure that the algorithm can be tweaked. $\endgroup$ Jan 10, 2017 at 0:12
  • $\begingroup$ @NeilStrickland I will mention that it is possible to link to a search - like here. This makes it easier for others if they want to try the same search - all they have to do is to click on the link. (But I guess that after including all four links, not much space would be left for the rest of the comment due to character limit.) $\endgroup$ Jan 10, 2017 at 1:50
  • $\begingroup$ @NeilStrickland Thank you so much for your feedback, after I made a little tuning on my ranking formula, now it looks much better if you search for $\Omega S^3$. For more detail see my explanation in chat room. $\endgroup$
    – Wei Zhong
    Jan 10, 2017 at 16:06
  • $\begingroup$ @WeiZhong: I have put some further test cases in a separate answer. $\endgroup$ Jan 10, 2017 at 17:18
  • $\begingroup$ @NeilStrickland I do not have enough reputation to comment under your answer, but like you have assumed in that answer, perhaps there is no relevant document there. So the best way to indicate there may be something wrong is to identify a post (ensure it is not too new, if so it may not be indexed yet) on MSE and you are unable to search it via Approach0. In this case, please report to me. $\endgroup$
    – Wei Zhong
    Jan 11, 2017 at 13:28
  • $\begingroup$ @NeilStrickland For your fourth point, you can get results by removing the a pair of bracket: approach0.xyz/search/… $\endgroup$
    – Wei Zhong
    Jan 19, 2017 at 14:01
2
$\begingroup$

It seems that uniquation searches both on MathOverflow and Mathematics Stackexchange (and also other sites).

Among example queries, there is search for $F_{n+1}=F_n+F_{n-1}$. You can see that among the results there are posts from Math.SE and posts from MathOverflow.

As I am not familiar with this search engine, I am making this post CW - feel free to edit this answer if you know more about uniquation.

$\endgroup$
2
  • 1
    $\begingroup$ Specifically, if somebody knows whether that project is abandoned or still being developed, it would be useful information. If it only indexes posts from MO which are at least 5 years old, it will probably not be too useful. (When I click on news on the uniquation website, I am directed to twitter account where the last tweet is from 2011.) $\endgroup$ Jan 8, 2017 at 19:09
  • 1
    $\begingroup$ It seems that this project has been abandoned. (Or possibly moved to another website?) When I try the link uniquation.com/en (which worked when I posted this answer), I get a page saying: "This Domain Name is Possibly For Sale." $\endgroup$ Oct 18, 2018 at 8:12
2
$\begingroup$

A few years ago we developed the search engine SearchOnMath, in order to search for mathematical formulas. Recently our tool has indexed both: Mathematics and MathOverflow.

Currently, SearchOnMath is the mathematical search engine with the largest number of indexed sites (including Wikipedia, Wolfram MathWorld, among others ...).

The following video illustrates how it works: SearchOnMath - a brief guide.

P.S.: please, enclose formulas between \${}\$. e.g. \${x+y}\$.

We’d love to hear your feedback.

$\endgroup$
1
  • $\begingroup$ I like it, just don't get why we need double delimiter \${}\$ instead of classic \$\$? This way it is a bit non-standard for anyone who is used to latex/mathjax. $\endgroup$
    – Sil
    Dec 18, 2019 at 0:22
1
$\begingroup$

Here are some further test cases. Some of them return nothing of interest from MSE. This may be because there really isn't anything relevant there.

  1. The Eilenberg-MacLane space $K(\mathbb{Z},3)$: link
  2. The ring spectrum $MO\langle 8\rangle$, entered with angle brackets or less than/greater than signs.
  3. The classifying space $BGL_n(R)$: link (The first result is good, most of the others are for $PGL_n(R)$, which is mathematically quite a different object.)
  4. The absolute Galois group $Gal(\overline{\mathbb{Q}}/\mathbb{Q})$: link. This one got no hits at all, which I found surprising.
  5. The cohomology ring $H^*(\mathbb{R}P^\infty)$: first attempt, second attempt. The first time, the input system converted the superscript * to \cdot, which is often reasonable but not in this context. That gave me no hits. I switched to raw mode and used a * for the second attempt, which is much better. After realizing this, I tried 4 again in raw mode, again with better results.
$\endgroup$
2
  • $\begingroup$ Ad 4., it is caused by the redundant $\{ \}$ brackets in "numerator", without that it seems to work fine. Although I think the search engine should be able to deal with it by itself. $\endgroup$
    – user124222
    Jun 20, 2018 at 7:08
  • $\begingroup$ When I try \operatorname{Gal} rather than just Gal and \mathbb Q rather then just Q (and also replace the fraction by /), then I get better results for 4: link. Although it is clear that this application has evolved a lot over the years, which might also influence what I get. $\endgroup$ Sep 10, 2020 at 5:32
0
$\begingroup$

SearchOnMath searches Math World and Wikipedia. They say you can contact them to get a site added.

$\endgroup$
1

You must log in to answer this question.