Quotation Source

Wikipedia: wikipedia article length etc. as metric of importance of topic

2007-10-09T20:53:58.0000Z

Yes, inlinks are a simple and useful measure. Another which has received attention is number of edits. e.g. Wilkinson & Huberman, Assessing the value of cooperation in Wikipedia. Because Wikipedia is large and well-documented, and there's general interest in large graphs and the way they scale etc., there have been loads of measures made of it for various purposes - e.g. see Voss, Measuring Wikipedia, Buriol et al, Temporal Analysis of the Wikigraph, Chernov et al, Extracting Semantic Relationships between Wikipedia Categories, Ollivier et al, Finding Related Pages Using Green Measures: The Example of Wikipedia. (But this is no doubt telling my grandmother to suck eggs, since it's apparently not just that Microsoft is interested in leveraging wikipedia for named entity disambiguation, but also that you guys have some connection with Powerset!)

Wikipedia: wikipedia article length etc. as metric of importance of topic

2007-10-09T19:24:12.0007Z

Hi there. Interesting idea (and thanks for sharing). I'd like to point out that Wikipedia has standards that require long articles to be split into multiple shorter articles. So at least for the more edited and popular topics, the length of articles all approach the same over time. On the other hand, each time an article is split out, a link is added to the main article to it, so the number of links would be a better metric of article "importance", if such a thing can be considered in such absolute terms.

Wikipedia: wikipedia article length etc. as metric of importance of topic

2007-09-28T14:32:01.0000Z

Sure, using these things to do autocomplete/predictive disambiguation requires dynamic calculations with local context (Though they usually of course work with some static global measure as their background.) But global measures of 'overall importance' like length of a wikipedia page, or PageRank for a web page, do tell you something informative for some purposes. E.g., it would be interesting to be able to compare freebase coverage to that of wikipedia: which pages within a given wikipedia category or freebase type have quite a lot said about them in wikipedia but not much about them in freebase? which of two wikipedia categories has better coverage in freebase? Again, if an app goes to freebase to retrieve information which is visually displayed it need not use any importance metric to select the information it wants but could use the metric in the display (size of text in a tag cloud, size of circle in a map) etc.

Wikipedia: wikipedia article length etc. as metric of importance of topic

2007-09-27T17:04:04.0000Z

That's a very interesting idea. We have an infrastructure internally that knows about the interconnectedness of wikipedia articles, it's part of the system that means that autocomplete and search know that "London" is more likely to be the one in England than the one in Ontario.

The issue with these metrics is that they are fairly dynamic and rather opaque and context sensitive. Different "George Bushes" are important if you are only interested in cricket rather than US politics for example. It's hard to get a single number which ranks "overall importance" in the same way that the web doesn't have a "most important page"; it all depends on what you are trying to do.

So I'm curious what applications you see as using this number (or set of data) and how they would use it.

Wikipedia: wikipedia article length etc. as metric of importance of topic

2007-09-21T17:10:20.0012Z

I'm reposting a comment I left at http://www.freebase.com/view/wikipedia, since no one seems to have noticed it there.

Lots of topics have keys in /wikipedia/en. I wonder whether it would be worth uploading, for each topic in this namespace, a simple measure of the relative importance of this topic in wikipedia. The simplest such measure would be the length of the wikipedia article (other obvious candidates: number of inlinks, number of outlinks, number of editors - even computation of the page's pagerank in the wikipedia graph, if the matrix computation was thought worth the effort.) This might allow more important topics to be selected for review to be given more attention: ceteris paribus, the more important a wikipedia topic, the more needs to be said about it on freebase. (I don't know if you have thought about metrics to measure how much is said about a topic in freebase.)

The Adventures of Buckaroo Banzai Across the 8th Dimension: Add a new quotation?

2007-08-29T03:02:39.0007Z

It is a little confusing. The combo box is generally how new objects, including quotations, are created. It may seem a little odd at first that a quotation is a full-fledged topic, rather than just a string-value somewhere, but a quotation is very legitimately something to which people might want to add other properties. Take a look at the Quotation description for some further information. If the quote you want to add isn’t already in the system, go ahead and create it (“Create New Quotation”), using those guidelines.

“Where are we going?”
“Planet Ten!”
“When?”
“Real soon!”

The Adventures of Buckaroo Banzai Across the 8th Dimension: Add a new quotation?

2007-08-28T23:39:21.0005Z

Hi - I'm still learning my way around freebase, and I wanted to add a quotation to this page, however I only appear to be able to select from a combobox. Am I not allowed to enter free form text there? Any guidance appreciated.

Thanks!
Iniko

Dan Bolser: How vain am I!

2007-08-24T15:04:30.0005Z

Very!

Goldfinger: split

2007-07-16T04:41:26.0011Z

the wikipedia article is for the film

Atlas Shrugged: Will it ever be made into a movie?

2007-05-03T15:09:24.0012Z

http://imdb.com/title/tt0480239/

Latest rumors are Angelina Jolie for Dagny and Brad Pitt for Galt.

Annie Hall: Excellent Movie

2006-12-20T15:27:04.0018Z

This and Manhattan are his best two films in my opinion