Comments on: Techzingo: news by tag http://libraryclips.blogsome.com/2006/02/22/techzingo-news-by-tag/ sharing ideas thoughts and feedback Sat, 28 Nov 2009 14:57:14 +0000 http://wordpress.org/?v=1.5.1-alpha by: Johnt http://libraryclips.blogsome.com/2006/02/22/techzingo-news-by-tag/#comment-24156 Mon, 13 Mar 2006 08:22:11 +0000 http://libraryclips.blogsome.com/2006/02/22/techzingo-news-by-tag/#comment-24156 Ted, I did forget to mention that even though Personal Bee works on text analysis, like TagCloud (auto-tags), there is quite a difference under the bonnet in the way of bubbling up unique topics. A lot of the times unique terms, or should I say phrases, are unique because they haven't been heard or strung together yet, meaning they must be new, and perhaps the latest breaking news. I believe Personal Bee tries to tease out these phrases...so it can be seen as a way to read fresh content, rather than a way to categorise content in general. I have quoted you guys in an <a href="http://libraryclips.blogsome.com/2005/09/05/personal-bee-relieve-feed-overload/" rel="nofollow">earlier post</a>: So how is it different to TagCloud, well for starters it is an RSS reader, but how is the keyword extraction different…here is the unique method according to the new instruction, HOW-TO Be a Beekeeper: “Each Bee edition ranks topics (i.e. keyphrases) proportional to their popularity within the interest window and inversely proportionally to their historical popularity. It’s a simple concept, yet it works well. Consider for instance the topic “Google.” Without an established historical baseline, the topic “Google” would constantly rank high on the topic list, thus pushing down other potentially new and interesting topics. In a Bee edition, “Google” would rarely make the topic list because of its high historical popularity. Instead, the day Google announced “Google Talk,” that topic immediately reached the top of the ranked topic list because the phrase “Google Talk” was mentioned 100’s of times and had no history in the edition.” …And since a popular topic requires multiple mentions before its ranking increases, it can take several days for interesting topics to “bubble” to the top of the list. As a corollary, topics at the top of the list can take several days to “fade away.” …The Personal Bee tracks multi-word keyphrases rather than single keywords. Single keywords are simply too general and/or ambiguous So the difference is that it seems to be effective in teasing out keyphrases for new unique terms, which is great for current awareness. Ted,

I did forget to mention that even though Personal Bee works on text analysis, like TagCloud (auto-tags), there is quite a difference under the bonnet in the way of bubbling up unique topics.

A lot of the times unique terms, or should I say phrases, are unique because they haven’t been heard or strung together yet, meaning they must be new, and perhaps the latest breaking news.

I believe Personal Bee tries to tease out these phrases…so it can be seen as a way to read fresh content, rather than a way to categorise content in general.

I have quoted you guys in an earlier post:

So how is it different to TagCloud, well for starters it is an RSS reader, but how is the keyword extraction different…here is the unique method according to the new instruction, HOW-TO Be a Beekeeper:

“Each Bee edition ranks topics (i.e. keyphrases) proportional to their popularity within the interest window and inversely proportionally to their historical popularity. It’s a simple concept, yet it works well. Consider for instance the topic “Google.” Without an established historical baseline, the topic “Google” would constantly rank high on the topic list, thus pushing down other potentially new and interesting topics. In a Bee edition, “Google” would rarely make the topic list because of its high historical popularity. Instead, the day Google announced “Google Talk,” that topic immediately reached the top of the ranked topic list because the phrase “Google Talk” was mentioned 100’s of times and had no history in the edition.”

…And since a popular topic requires multiple mentions before its ranking increases, it can take several days for interesting topics to “bubble” to the top of the list. As a corollary, topics at the top of the list can take several days to “fade away.”

…The Personal Bee tracks multi-word keyphrases rather than single keywords. Single keywords are simply too general and/or ambiguous

So the difference is that it seems to be effective in teasing out keyphrases for new unique terms, which is great for current awareness.

]]> by: Ted Shelton http://libraryclips.blogsome.com/2006/02/22/techzingo-news-by-tag/#comment-23690 Fri, 10 Mar 2006 16:38:50 +0000 http://libraryclips.blogsome.com/2006/02/22/techzingo-news-by-tag/#comment-23690 Thanks for mentioning the Personal Bee - although we are quite ready for the attention! Still working to get our beta ready to launch later this month. But I thought I'd comment on one point that you made here -- the "cloud" that the Personal Bee generates is <strong>not</strong> a tag cloud. Tags are user generated labels that may or may not be good indicators of what something is about and do not evolve over time... by contrast what we are doing (which we call a <em>phrase map</em>) is extract actual phrases being used in the articles in a given topical group (which we call a Bee), analyze them for time relevance, and present them ranked according to a relevance algorithm that we have developed which takes into account usage frequency, distribution across feeds, interest window frequency, and other factors. So we believe that a phrase map gives you a window into what a set of feeds are talking about right now -- as opposed to a static view imposed by a set of historical tags. Best - Ted Shelton, CEO The Personal Bee, Inc. Thanks for mentioning the Personal Bee - although we are quite ready for the attention! Still working to get our beta ready to launch later this month. But I thought I’d comment on one point that you made here — the “cloud” that the Personal Bee generates is not a tag cloud. Tags are user generated labels that may or may not be good indicators of what something is about and do not evolve over time… by contrast what we are doing (which we call a phrase map) is extract actual phrases being used in the articles in a given topical group (which we call a Bee), analyze them for time relevance, and present them ranked according to a relevance algorithm that we have developed which takes into account usage frequency, distribution across feeds, interest window frequency, and other factors. So we believe that a phrase map gives you a window into what a set of feeds are talking about right now — as opposed to a static view imposed by a set of historical tags. Best - Ted Shelton, CEO The Personal Bee, Inc.

]]>