Take the 2-minute tour ×
Ask Patents is a question and answer site for people interested in improving and participating in the patent system. It's 100% free, no registration required.

AN OVERBROAD PATENT ON performing frequency analysis on the occurrence of target words in a stream of words - This application from Citizennet seeks to patent the idea of...Generating a discrete time sequence of counts for a target word, performing frequency analysis of the generated discrete time sequence, and determining that the target keyword is a trending keyword based on the frequency analysis! 10 minutes of your time can help narrow US patent applications before they become patents. Follow @askpatents on twitter to help.

QUESTION - Have you seen anything that was published before 8/7/2011 that discusses:

  • Detecting trends from a stream of words based on frequency analysis of word counts

If so, please submit evidence of prior art as an answer to this question. We welcome multiple answers from the same individual.

EXTRA CREDIT - Streams of words are taken from word sources such as news stories, tweets, posts on social networking sites, or blog entries.

TITLE: Determine whether a target word is a trending keyword based on frequency analysis of discrete time sequence of counts for the target word

Summary: [Translated from Legalese into English] Generating a discrete time sequence of counts for a target word from a stream of words, performing frequency analysis of the generated discrete time sequence to determine contributions of frequency components within different frequency ranges, and determining that the target keyword is a trending keyword based on the frequency analysis.

  • Publication Number: US 20130036107 A1
  • Application Number: US 13/568,851
  • Assignee: Citizennet
  • Prior Art Date: Seeking prior Art predating 8/7/2011
  • Link to Google Prior Art Search - "Find Prior Art"

Claim 1 requires each and every step below:

A method for trend detection, comprising:

  1. Generating a discrete time sequence of word counts for a target word using a trend detection device;

  2. Performing frequency analysis of the discrete time sequence of word counts to determine contributions of frequency components within different frequency ranges to the discrete time sequence of word counts using the trend detection device; and

  3. Detecting that the target word is a trending keyword based upon at least the frequency analysis of the discrete time sequence of word counts for the target word using the trend detection device.

In English this means:

A method for trend detection, comprising:

  1. Generating a discrete time sequence of counts for a target word;

  2. Performing frequency analysis of the generated discrete time sequence to measure frequency of target word counts within different frequency ranges; and

  3. Detecting that the target word is a trending keyword based on the frequency analysis.

Good prior art would be evidence of a system that did each and every one of these steps prior to 8/7/2011

You're probably aware of ten pieces of art that meet this criteria already... separately, the applicant is claiming Performing frequency analysis refers to applying discrete Fourier transform to the discrete time sequence of counts for the target word(s).


"Trend detection using frequency analysis of streams of words" from the Applicant


What is good prior art? Please see our FAQ.

Want to help? Please vote or comment on submissions below. We welcome you to post your own request for prior art on other questionable US Patent Applications.


share|improve this question
1  
In terms of the first parts, tracking the frequency of keywords, that has been widely available for a long time. We had that in Joomla 1.0. I have to look around a bit in terms of part 2 (time limited) and part 3 (identifying as trending within a given time). –  Elin Jan 21 at 23:21
    
I think the main idea is to take the time-series of keyword counts, do a Fast Fourier Transform on the time-series, and look for patterns in the frequency domain which are indicative of a keyword trend. The claims are of course written more broadly than whatever specific algorithm they have developed for predicting keyword trends, which is part of the problem. Performing Frequency Analysis is a very general idea. I would rather see the claim limited to a highly specific form of frequency analysis, hopefully focusing on specific non-obvious features which predict trends in keywords. –  Micah Siegel Jan 21 at 23:27
    
It wasn't clear to me that they were predicting; it seemed more simple and retrospective. Frequency is definitely very general. –  Elin Jan 21 at 23:58
    
Actually, this is any stream of words, right? So this could be the data you get that tells you what search terms are bringing people to your web page? –  Elin Jan 22 at 0:02
    
Ideally prior art would something which has performed each step in the method (Generating time sequence..., Performing frequency analysis..., Detecting that word is trending...) –  Micah Siegel Jan 22 at 0:08

3 Answers 3

This may or may not apply.

http://www.esaim-ps.org/action/displayAbstract?fromPage=online&aid=8211879

ESAIM: Probability and Statistics / Volume 13 / July 2009, pp 328-342

share|improve this answer

An example of a common situation where people want to use streams of words and analyze their frequencies and how those frequencies change over time is the analysis of search terms that people use to find a website (i.e. If I went to Google or Bing and searched for "patent discussion" and then clicked on the result, the site I went to would want to know those are the terms I searched rather than say "patent debate"). In the search engine optimization industry this is known as organic search and traditionally the search engines have provided this information via the http request.

As far back as 2005 Webalizer (an open source application for analyzing server logs) allowed the analysis of frequency of search terms by time. There are many forks of Webalizer some of which since 2005 have further enhanced this aspect of the analysis of the logs (e.g. http://ostermiller.org/webalizer/ http://www.stonesteps.ca/). See also http://www.daydream.co.uk/webalizer_guide.asp#webalizer_top_and_total_search_strings -- the whole page discusses the use of Webalizer to identify trends.

Awstats is another application for doing this. There are many extensions to Awstats that provide more complex analyses of trending search terms.

share|improve this answer

Would Wordle be in the right direction as prior art, or is this not specific enough? http://www.wordle.net/ There are examples on the Wordle website which are 5+ years old.

I also found a blog from 2009 which contextually appears the same as "Detecting trends from a stream of words based on frequency analysis of word counts". The blog author states:

I used Wordle to make the "word cloud" below out of the text from President Obama's speech to Congress. I chose a setting to display the 25 most frequently used words in his speech. Glad to see that education made the top 25 of his verbal agenda!

http://www.peterpappas.com/2009/02/president-obamas-address-to-congress-22409-education-in-the-top-25.html

Not sure about the 'Generating time sequence' step mind you. Any ideas how a 'Generating time sequence' would appear in prior art?

Hope this helps a bit

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.