Pandia Post No. 22 Part 1

News search engine Topix.net gets even better

By Lars Våge

Recently the news search service Topix.net was upgraded with several new features like email alerts and (according to them) better relevance ranking.

(August 2004) Topix.net first appeared in March 2003 and among the people behind it are some of the founders of the web directory Open Directory Project, among them ODP creator Rich Skrenta.

Topix is short for topics and this is the main thing in this news search service. But maybe it would be more properly called a news aggregator: on the front page you will find much more than a search box. What you find looks more like Google News, RocketNews, Columbia Newsblaster and other news sites that will sort the news into different categories and present them in a way that is similar to news sites with human editors.

What makes Topix.net so excellent and the reason it is so appropriately named is that the automatic categorizing of news is taken to a whole new level. They offer a whopping 150 000 news categories: a category for every city and every public company in the United States, categories for 1 500 teams in different team sports, 48 000 celebrities and much more.

Compare this to the 7 news categories on Google News, 8 on Rocket News, about 100 on Yahoo News and some 700 on NewsNow.

The news is selected from 7 000 different news sources. They are for the most part American, which is a drawback for users elsewhere. None the less there is much of interest to be found on Topix.net.

Every category of news has a RSS version that you can subscribe to and read in a RSS reader. Now you can even receive daily or weekly email alerts with the most recent news on a specific topic.

[RSS is a text file (XML) format used for syndicating or distributing headlines and content from Web sites. You can use programs or sites that reads these RSS-files to check the feed for new news items and other changes. The Editor.]

To get the address of the RSS version you click on the XML icon to the left under the heading Tools. This is also where you find the link ‘Get news alerts’ if you want news alert emails.

You can’t get email alerts for your own searches. For this you will have to be content with the RSS version. You get the RSS version of your search by copying the address behind the orange XML icon on the left hand side on the result page.

How do they go about placing the news in all these thousands of categories? First, every news article is analyzed to extract geographical references that are compared to a geographic database to be assigned a geographic category.

Next, the content is analyzed to be matched to the 150 000 concepts defined in the Topix.net Knowledge Base which utilizes 10 million language patterns in the categorization process.

So every news article is categorized with regard to both geographic location and topic. Different versions of the same story are sorted out so you will find no duplicates in the results. If you want to see every available report on a certain news event, click on “Full coverage” to see the other articles.

In addition to the categorization, every article is also assigned a NewsRank, depending on how important it is. This is decided after taking into account 17 different factors.

Topix.net has undoubtedly raised the stakes and Google, Yahoo, and even Microsoft (who have the beta version of their Newsbot out) will have to watch out. None of them are even close to the kind of advanced categorization of news that Topix.net offers.

Go to page 2: On the Kozoru natural language search engine