Feature request - the "way out there" list

Give him an inch and he'll take a mile...

Yes, we know we were super lucky with our timing to join and to suggest group lib file saving ...and then a week or so latter, here it is. We understand it must have been in the works for sometime, but we still feel blessed.

So why not try some other ideas? :-)

Retrieve Tags.

I love the tag function, but it is a bit tedious to input as is. Ideally, I'd love to see a retrieve tags function, just like the retrieve metadata function (which is uber cool, by the way!) Retrieve Tags could even be a pan-collection option, where a search was done across the entire collection to find ALL items that contained such keywords and recorded such tags in the metafile....

or the function might suggest tags from say keywords that are sometimes listed in article or from metadata tags in the source...or some algorithm that selects rare nouns from the abstracts.

Retrieve Related Items:

And a retrieve related items could be done in a similar manner. The same or similar collection wide keyword search for adding tags could be used to say link an article with any other articles that reference it in their bibs. And the uber cool function here...way out on our wish list.... an XML like list of such up and down references of related items added to an auto spawned Report page for each item with related items.

Such related item info could then be used to generate a cool spider wed graphic of the relations in an entire library. With the right GUI, allowing cross collection item selection, it could lead to a whole ease of doing research.

And another? Common references found by in say two or more bibs of different items, but missing from a library, could be added to a Suggested Readings file, noting the ref sources.

Obviously, a lot of coding...and I'm found I was bad at coding back in the 90s.
  • you can do this already for tags: you can search for items (maybe even use a saved search) and then tag them all at once with one tag by selecting them and dragging them onto the tag.
  • I know one user who has written a really cool data-mining tool to extract terms (especially rare nouns and noun combos) from the full-text and create a lovely an informative index.
    It's an early alpha, and needs some more input - but I hope he gets it out there soon.
    It's certainly "über cool" - the potential is HUGE!

    I can sort-of see how the other parts might work, extracting "titles" from the text and cross checking the existing ones - though it would surely be intensive, and error prone (e.g. for short titles).

    The long-term potential is amazing. I'm very excited to see where this all goes in the next 3-10 years!
  • Hey adamsmith,

    The way you explained it would be perfect....just on problem...I'm not sure it works. At least not yet.

    I tried as I thought you stated, without love.

    Did a search (simple and advanced), had tag open for the search term in the right pane, selected all files with the intent to drag onto this tag... but as soon a I select more than one file, the right pane grays out to "# items selected"...not allowing me to drag all on.

    Care to walk us through your tag patch in more detail?
  • You drag to the tag selector in the bottom left.
  • http://www.zotero.org/support/screencast_tutorials/tags
    towards the end of the screencast
  • Ahhh...the hidden tag pane. Forgot about that.
    Got it. Thx.
  • The user komrade referrs to is me. It's a locally running perl web application that runs on top of an existing zotero database and storage directory. It provides a list of key phrases and named entities, a "rating" scheme, and links back to the document corpus.

    It's written in Perl with Catalyst (http://catalystframework.org), and uses the Template Toolkit for data display.

    Unfortunately I developed it for zotero 1.0 and the database structure is slightly different, so the document corpus doesn't index as well. I made some integration with the opencalais web service (http://www.opencalais.com/ a thomposon reuters product, the irony) for named entity extraction as well as indexing the documents at the phrase level.

    Things that ought to be on the roadmap are restricting by collection and by date added, but I don't have the resources to do so yet. Anyone with any Perl chops who wants to have a look at this, contact me via http://www.uow.edu.au/~kd21 and I'll put it up on github.
  • Interesting...thanks for the add. I just checked back here as I came across and interesting article related to mapping such meta data.

    ANALYZING AUTHORS AND ARTICLES USING KEYWORD EXTRACTION,
    SELF-ORGANIZING MAP AND GRAPH ALGORITHMS

    http://www.cis.hut.fi/tvatanen/sompa.pdf

    Seems they are already using with BibTeX...

    Such a map, and ideally generating suggested new content alerts/feeds for Zotero would be a uber Killer ap!
Sign In or Register to comment.