ABOUT YOUR LISTENING POST™

What data is being collected?

Any relevant blogs that can be found using the Google, Blogger, and Technorati search engines.  Other sources may be monitored, per subscription options.  Relevant sites are found via conventional key word search.  You can specify search terms which are used to initiate the search, and inclusion terms which ensure that you get only the posts that are truly relevant.

How are the results analyzed?

Data is either presented per last week and last 30 days, or per last month and last 30 weeks.  Graphs all presented per last 30 days.  The key metrics are

• Low relevance articles—Articles in which search terms have insignificant importance
• High relevance articles—Articles in which search terms have significant importance
• Tone—Fraction of text that is positive or negative sentiment.  This is measured using dictionary of 20,000 words and can range from from -1.0 to +1.0, although it rarely reaches those extremes.  Across many types of web content, tone tends to average around 0.10 (i.e. 10% positive).

From	To	Tone
-1.0	-0.3	Very negative
-0.3	-0.1	Negative
-0.1	0.0	Slightly Negative
0.0	0.1	Slightly Positive
0.1	0.3	Positive
0.3	1.0	Very Positive

• Focus—Fraction of text that refers back to itself, i.e. its degree of coherence.  High focus means a few topics discussed coherently, while low focus means many topics discussed casually, as in casual conversations. Positive tone is often associated with low focus, while negative tone and high focus can be a warning of dissatisfaction.

From	To	Focus
0.0	0.1	Low
0.1	0.4	Average
0.4	1.0	High

• Significant words fall into three classes.  Hot words  are the most important words found in the articles.  Importance measured using natural language processing.  Happy words are words that tend to be hot in positive-tone articles.  Grumpy words  are words that tend to be hot in negative-tone articles.

Interpreting the analysis

Listening Post™ is an exploratory tool which you can use to validate and enrich existing knowledge, challenge existing knowledge, and discover new knowledge.  

• Look at the aggregate first (don’t over-interpret when the number of articles is very low).  Are there any changes in tone, focus, or volume?  Do these changes correspond to events we know about?
• What are the key words (hot, happy, grumpy)?  What might they imply?  Do they collectively suggest any themes?
• Look at individual entries.  To select “interesting ones” you can use titles, the README column, or ones with very low or very high tone, high focus, high relevancy, or high mentions (the number of times the search term(s) occur in the article.

If you have additional questions please email us.