My company has just released a new service called ContentSieve (http://www.contentsieve.com). For further information please have a look at the following press release, taken from http://metasieve.wordpress.com/:

More and more information providers offer their content on the Web. In order to be successful they have to attract a large number of users to their site. There are two main ways of achieving this. One can either advertise on other sites or once can try to improve the site’s search engine listing through optimisation of site and link structure.

Attaining a reasonably good listing in the usual search engines becomes increasingly difficult. There is intense competition for each search term. Many companies hire professional search engine optimisers for gaining a prominent position in search engine results. Each year millions of euros are spent on this.

Smaller competitors increasingly fail to participate in this market. In the meantime the best search engine optimisers charge up to a whopping 1,000 EUR per month. The prices for creation of content that is considered relevant by search engines keep climbing with no end in sight. Even interesting information frequently is only displayed by search engines if it complies with the complicated rules set up by search engines. This drives up the cost of content creation and creates market entry barriers which hardly any small competitor can handle.

ContentSieve is a new online service helping Web content providers to overcome these barriers.

"As of lately, it has become more difficult to differentiate one’s Web site in order to capitalise on its content." says Björn Wilmsmann, CEO of MetaSieve, the provider of the new service. "High costs for editorial content, AdWords and search engine optimisation leave the online content business to but a few large competitors."

With ContentSieve online service providers now can add up-to-date and thematically matching content to their sites. This way, news flashes or search results for a specific topic can be displayed. This content is integrated as widgets into existing sites. Permanently updated content makes a Web site more attractive to users and brings about a better search engine ranking. The content is already optimised so there is no need for expensive search engine optimisation anymore.

"Our goal," explains Wilmsmann "is to offer access to attractive online marketing opportunities, especially to small content providers."
ContentSieve is ready for use after just a few clicks. After having registered, the user can create widgets and include them in her own site with a small code snippet. Subsequent editing of this code snippet is not needed anymore as third party content can be delivered through this code as well. Billing works through ordering packages, which contain a certain number of widget views at a fixed price.
ContentSieve immediately begins to deliver relevant content until these widget views are used up. To allow extensive testing each new user will get 5,000 widget views for free as a welcome gift.

ContentSieve currently is available in English and German.

Today my company MetaSieve has released a new online service called GameSear.ch. GameSear.ch is a new vertical search engine that aggregates a huge number of sources about games - computer and video games in particular - all in one place.

GameSear.ch allows you to find, access and syndicate all the relevant gaming content on the web.

I have just found out about a very cool project that makes use of the Topicalizer API called FeedBlast. The detailed concept is described under http://www.rndparty.com/

The basic idea is to take RSS feeds, have them analysed via Topicalizer and to create an audio / video summary from the input. For this purpose an AT&T text-to-speech software will be used to read out the abstract created by Topicalizer. Moreover, the keywords provided by Topicalizer are used to find media items that fit the RSS content.

Very cool stuff, indeed. I am looking forward to hearing more of this.

For those of you with sufficient command of the Spanish language: Topicalizer has recently been covered in the Spanish language weblog 'La Bitácora del Tigre' (The Tiger's Log): http://www.labitacoradeltigre.com/2006/12/23/mas-recursos-de-escritura-y-feliz-navidad-para-todos/
Eduardo-Martín Larequi Garcí, the author of this weblog, is a Spanish language and literature teacher, who is currently employed by the Department of Education of the regional government of Navarra.

Oh, before closing, I would like to wish you a Merry Christmas (or 'Feliz Navidad' for that matter).

There is a new tool on this site, which I have called Semantic Web Creator

This tool creates a set of related terms for a given term. In order to achieve this, it makes use of the Google API and Wikipedia.

The Google API is used to get the Wikipedia page for the term and this page is scanned for links to other Wikipedia articles in order to retrieve the terms which are associated with the original term.

Bookalizer.com (http://www.bookalizer.com/) is a web service that allows you to display customised Amazon book ads on your homepage according to the content of the respective page. You can either specify the key words by which books are selected by yourself, or provide a set of Amazon product IDs or have Topicalizer select the most appropriate key words automatically.

The service is organised by a German organisation called Clever Spenden e.V., which donates the revenues of affiliate services to a good cause.

Please have a look at their weblog as well: http://bookadmin.blogspot.com/

Topicalizer now also calculates tetragram and pentagram results (i.e. four- and five-word phrases) besides the bigram and trigram ones. Furthermore, Topicalizer from now on does not only display the ten most frequent results for each category, but the overall most frequent results as well.

The new tool available under the URL http://www.topicalizer.com/coOccurrences/ tries to guess the words that are most likely to co-occur with the words you entered. That is, if you enter 'president' for example, the software will offer you, amongst others, the word 'kennedy' as possibly occuring in the same context.
This tool draws upon the Brown Corpus of Present-Day Edited American English.

This new feature is also part of the Topicalizer API.

Under http://www.topicalizer.com/augmentKeywords/ you can enter a sentence, phrase or word (you could possibly also enter longer texts, but the generated results would most probably be too large to make sense), whose keywords will be extracted along with their synonym / hyponym / hypernym relations by using WordNet.

This new feature is also part of the Topicalizer API.

There is a new screencast about Topicalizer and how it can be used on a blog called 'Sawfish Mango' (http://www.sawfishmango.com/), please have a look at http://www.sawfishmango.com/screencasts/mashup_paper.html for the screencast and http://sawfishmango.com/?p=19 for the related article.

Topicalizer now detects and displays custom document character sets, where before it simply assumed everything to be ISO-8859-1 or UTF-8. If no custom character set is detected or if plain text is submitted, UTF-8 is used instead.
This improved behaviour regarding character sets leads to an improvement concerning automatic language recognition and the overall analysis as well, especially for languages that contain many special characters like for instance, Norse languages or Russian.

Topicalizer is now listed at http://www.buzzshout.com/ and http://www.neobinaries.com/:
http://www.buzzshout.com/listing.php?name=Topicalizer
http://www.neobinaries.com/apps/Category/Miscellaneous/Topicalizer.aspx

Buzzshout and NEO Binaries both are websites which present new products related to what is now commonly known as 'Web 2.0' and have them reviewed by their community.

There was a bug that prevented stopwords from being recognised correctly when occuring right before a sentence boundary. This issue has just been solved, so stopword recognition should be even more reliable.

Today I have fixed a bug that occasionally led to JavaScript and CSS not being filtered before analysing an HTML document. Now JavaScript and CSS elements are removed correctly before the actual analysis is conducted, so they do not falsify the results anymore.