search

McGill-NLP / mcgill-nlp.github.io

https://mcgill-nlp.github.io
0 stars 16 forks source link

Improve Google SEO, author and article metadata #372

Open AndreasMadsen opened 3 days ago

AndreasMadsen commented 3 days ago

Would be cool to have profile pages with this kind of metadata and

And link it with the paper pages using:

I can see there is already some metadata. However, I think it's incorrect. For example the authors are currently structured as:

<meta itemprop="description" content="Andreas Madsen, Sarath Chandar, Siva Reddy">

it's also listed as a CreativeWork which I don't think google supports. It should properly just be Article.

I think you can setup Google Search Console https://search.google.com/search-console/about to better inspect what google sees.

xhluca commented 2 days ago

Good idea! Do you know where we could we start (in terms of changing the jekyll template)?

AndreasMadsen commented 2 days ago

For papers:

  1. Restructure authors to be a list of strings instead of a string, e.g.: https://github.com/McGill-NLP/mcgill-nlp.github.io/blob/87da2660d831e2174a8656d77f19026a8c836a59/_posts/papers/2024-07-21-2310.07819.md?plain=1#L4
  2. I can't find the Jekyll template for papers, but use https://developers.google.com/search/docs/appearance/structured-data/article and refer to authors by a given ID

For authors: This is complicated because of the customizability

  1. Maybe complete https://github.com/McGill-NLP/mcgill-nlp.github.io/issues/330
  2. Use https://developers.google.com/search/docs/appearance/structured-data/profile-page
AndreasMadsen commented 2 days ago

https://developers.google.com/search/docs/appearance/structured-data/organization is properly also a good idea. Could start with that, since it's simple.

xhluca commented 2 days ago

Thank you. This is a very good idea. Would you have the bandwidth to create a PR for this? @cesare-spinoso and I would be happy to review it.

AndreasMadsen commented 2 days ago

I already tried to look at it. However, I don't understand how the generation actually flows. For example, I cannot find where <meta itemprop="description" content="Andreas Madsen, Sarath Chandar, Siva Reddy"> comes from. Perhaps it's something deep within minimal-mistakes. So I don't think I will be implementing this now.

xhluca commented 1 day ago

Could it be this? https://github.com/mmistakes/minimal-mistakes/blob/6be43f77a50598643e092117b50c0ea72ad9c94a/_layouts/single.html#L26

AndreasMadsen commented 8 hours ago

Yes, that looks like it. To set it up correctly with authors, etc., I would have to mess with that setup. I don't want to start that.

I will try to setup something similar on https://chandar-lab.github.io, then you can look at that if you are interested. But we don't keep as much information there, e.g. article abstracts.

xhluca commented 8 hours ago

Awesome!