Comments on: Visualizing blog language data

By: Cornelius

Cornelius — Fri, 31 Aug 2007 21:31:54 +0000

Thanks again for making me aware of this and for the great exchange of ideas we had, Anjo. I felt that there was something funny about the scores and in retrospect it’s certainly something I should have fixed earlier.

From what I can see so far it actually doesn’t seem to effect the tendencies I’ve discovered before though – in most cases they are actually more pronounced now. I’ll recalculate my scores and post something on that soon…

By: Anjo

Anjo — Thu, 30 Aug 2007 20:30:57 +0000

Cornelius.

There might be a “small” problem with the computation of the F-score as given above and the F-score as defined by Heylighen and Dewaele (2002). You seem to use frequencies: ((6 nouns – 4 verbs) + 100) / 2 = 51 (F-score). However, [HD2002] work with percentages: ((60% nouns – 40% verbs) + 100) / 2 = 60 (F-score).

Whether this invalidates your results is difficult to say, but it may be clear that working with percentages scales for the size of a document, whereas working with frequencies does not. The [HD2002] has does been designed to keep the F-score between 0% and 100%.

Interestingly, with percentages one can also apply it to very short documents (such as chats).

Anjo.

By: What McDonald’s is saying - CorpBlawg

What McDonald’s is saying - CorpBlawg — Wed, 09 May 2007 08:30:23 +0000

[...] a) f-score (for details on what that is, read this post) [...]

By: iyliee

iyliee — Fri, 23 Feb 2007 17:56:20 +0000

interesting research there.
blogging to me is like a daily NEED.
its not about the things i do or what i want. but its more abt what i feel, what i want them to know about my life. and my blog to me is like whole story of life’s journey.

i see my blog there in one of ur sources. i was rather shocked actually. well do continue ur research and all the best.

iyliee.

By: CorpBlawg » A great conference that I’ll probably miss (ICWSM, Boulder)

Thu, 22 Feb 2007 16:28:19 +0000

[...] I must have buried my head too deep in the sand lately (the sand being my f-score stuff), because I somehow managed to overlook the whole event up to now. If I hadn’t I certainly would have submitted something there. [...]

By: CorpBlawg » Communication is complex

CorpBlawg » Communication is complex — Tue, 20 Feb 2007 07:31:22 +0000

[...] Krishna Kumar has written a great post summarizing what I’ve been doing lately with f-scores in blogs. My favorite quote: [...]

By: Krishna Kumar

Krishna Kumar — Sun, 11 Feb 2007 18:58:08 +0000

Just posted my comments at http://krishami.blogspot.com/2007/02/corporate-blogging-research.html. Your research is very fascinating and I wish all the best.

By: Cornelius

Cornelius — Fri, 09 Feb 2007 19:30:10 +0000

It tends to work like that, although it depends on many (!) factors, and a high or low score can mean a lot of different things. Also, one shouldn’t get the idea that a high score is “better” in any way. Lower scores are characteristic for more conversational or “involved” blogs. Simplifying a lot: low scores = more like a conversation, high scores = more like a ‘traditional’ written text.

By: Krishna Kumar

Krishna Kumar — Fri, 09 Feb 2007 15:32:08 +0000

Thank you very much, Cornelius. This is a very useful tool. I will be writing a review on this and will send you the URL. As you mention, the more frequent you post, the lower the f-score.

By: Cornelius

Cornelius — Fri, 09 Feb 2007 12:07:41 +0000

Done. Have a look here
http://services.alphaworks.ibm.com/manyeyes/view/SThgcEsOtha6C2-WBJadE2-