2

Are the complete set of Tags and links to corresponding questions publicly available?

If so, does anyone have experience mining them? Browsing vs Searching. Browsing could be improved by graphical ZUI (Zoomable User Interface) similar to Google Maps or Prezi (Flash based).

Zoom in to the rank-ordered histogram of Tags until links to individual questions are visible.

  • 1
    the tag system isn't hierarchical, so a zooming interface will have at most one level to dig. And the second level explodes in thousands of items – Dr. belisarius Dec 18 '14 at 20:38
  • Hierarchies and related indexes (eg posets/DAGs) could be constructed dynamically based by exploring relations b/w tags & Wolfram Language, like here: mathematica.stackexchange.com/questions/66397/… – alancalvitti Dec 18 '14 at 20:45
  • @belisarius, btw could be applied to other SE groups, some of which might be more hierarchical than others, eg Math. If the tags there can be related to AMS (3-tier) or similar classifications. – alancalvitti Dec 18 '14 at 20:47
  • That is a long shot. I don't think SE sites are worried about tag normalization. But perhaps I'm wrong. – Dr. belisarius Dec 18 '14 at 20:49
  • Depends what you mean by normal, eg translate.google's engine is based quantitative n-gram analysis of ~10^7 documents that were not manually normalized. You can do a lot w/ large datasets. And SE should worry about scalability as the Q&A corpus increases. It's already hard to search, and browsing is no better than wikipedia. – alancalvitti Dec 18 '14 at 21:00
  • SE was criticized many times about that. They always answer more or less the same: "searching is not our core business, use Google"- – Dr. belisarius Dec 18 '14 at 21:06
  • Google PageRank wasn't meant for graph (or tree etc) display & navigation, even though they own maps.google, a hierachial ZUI. You'd think they could connect the dots, eg display Google Drive folders by a more immersive tree view than last century's Finder model. – alancalvitti Dec 18 '14 at 21:15
  • I would like to see a useful browsing interface. I'm just saying that it ain't easy, and the powers that be are not concerned about facilitating it. – Dr. belisarius Dec 18 '14 at 21:22
  • 1
    @belisarius, another view could be by timeline, where tags are arranged in parallel "swim lanes" like this > mathematica.stackexchange.com/questions/60927/…. These can then be sorted vertically by various criteria like volume, most recent etc. Though much of the information will be iconized or suppressed, timeline views afford a more relational view vs tag-by-tag navigation. – alancalvitti Dec 20 '14 at 23:49
  • That's a very interesting idea – Dr. belisarius Dec 21 '14 at 1:20
1

I am not sure I understand but this page lists all tags by popularity (occurrence) and clicking on a specific tag name will bring up a (multi-page) list of all questions bearing that tag.

  • That page is the inspiration, but rather than clicking manually, is there a URI that exposes them all for Import? – alancalvitti Dec 18 '14 at 4:05
  • @alancalvitti I think you are looking for this: api.stackexchange.com -- please explore that yourself and tell me if that is not the case. – Mr.Wizard Dec 18 '14 at 4:06
  • That looks like work to review. Looking for something easy to get started like Import["m.se/tags/get=all&links=true","JSON"] almost 2015, the age is REST. Do you know linked data JSON-LD and Named Data Networking (NDN)? Everything is served as a resource w/ a global URL. – alancalvitti Dec 18 '14 at 19:13
  • @alancalvitti you will probably be interested in SE analytics by Pickett :) – Kuba Jan 23 '15 at 13:24
  • @Kuba, I tried to import all users, but longer lists, like pagesize=200, get "Bad Request". – alancalvitti Jan 24 '15 at 17:47

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .