Comments for Next Generation Media Monitoring

Comment on Named entity recognition: Paris Hilton, or the Hotel in Paris? by admin

admin — Tue, 06 Jul 2010 15:38:39 +0000

The tool works best within specific contexts. If the topic is too general, like the news categories you mention, there will be overlap and difficulty in creating a category with specific properties. Financial news is for example frequently peppered by political decision making, and financial institutions frequently sponsor sport events.

As for pricing, this is an analytical tool and not included in our monitoring price model you refer to. It is determined on a case-by-case basis, i.e. how much effort do we have to put into the machine learning process.

Comment on Named entity recognition: Paris Hilton, or the Hotel in Paris? by TomSan

TomSan — Tue, 06 Jul 2010 15:26:43 +0000

Can this be used for topical information too? News topic Sports, Finance, Politics etc.? Also, what’s the pricing, has it anything to do with your standard product offer? http://www.imooty.eu/product.php

Comment on Named entity recognition: Paris Hilton, or the Hotel in Paris? by admin

admin — Tue, 06 Jul 2010 14:35:51 +0000

For certain ontologies we can also determine the likelihood that an article belongs in one category or the other. For example, the above headline is really about Paris Hilton and her stay in the French capital and the software can assign a 60% probability that it belongs in the celebrity category.

Comment on Named entity recognition: Paris Hilton, or the Hotel in Paris? by admin

admin — Tue, 06 Jul 2010 14:35:25 +0000

There’s a feedback function so that the user can correct misplaced articles. This way the software can continually develop and improve.

Comment on Named entity recognition: Paris Hilton, or the Hotel in Paris? by Neilcro

Neilcro — Tue, 06 Jul 2010 14:24:39 +0000

What happens if a post falls in both categories, for example “Paris Hilton visits Paris, booked in at the Hilton”?

Comment on Named entity recognition: Paris Hilton, or the Hotel in Paris? by corneliapiper

corneliapiper — Tue, 06 Jul 2010 14:20:52 +0000

Sounds fantastic, I know a few people at Orange Mobile who will jump at this!

Comment on Social Media Monitoring: Why Google Alerts won’t cut it by admin

admin — Mon, 03 May 2010 21:51:56 +0000

Good question, thanks for asking it. There are two solutions.

The quick and dirty solution is by filtering information with AND and NOT denominators. For example “pirates” have to be mentioned in connection with “music”, “software” and “computer” and exclude post with “somalia”, “ship” and “highjacked”.

A more sophisticated way is using one of our semantics components which categorize data according to a machine learning / user feedback process. This is a tailor made service and generally for larger clients that need to categorize large amounts of data. An ontology is then created manually and after the initial machine learning process, the AI categorize information according to a fulltext ‘meaning’ analysis.

Comment on Social Media Monitoring: Why Google Alerts won’t cut it by jacksonwilliam0

jacksonwilliam0 — Mon, 03 May 2010 19:45:24 +0000

How do you deal with words such as “pirate”, won’t the IP lawer also have to deal with a lot of irrelevant information on ships around the Somali coast??

Comment on Social Media Monitoring: Why Google Alerts won’t cut it by alexwarren

alexwarren — Mon, 03 May 2010 18:10:39 +0000

I agree, the Google search parameters will make you a halfway psycho if you try using it professionally. In addition to the factors you’re mentioning, you’ll have to deal with some pretty steep language-, and possibly your IP location issues too.

Comment on Online reputation management, 2 to 12 at Goldman by admin

admin — Thu, 22 Apr 2010 15:55:38 +0000

Good point!