By: E&P Staff
Montreal-based Nstein Technologies Inc. has released Semantic Site Search (3S), “a front-end, multi-index search engine” that leverages Nstein text-mining technology to power a faceted site search for accurate, categorically organized results.
“Standard, statistical site search often doesn’t give users the results they are looking for,” Nstein Chief Technology Officer Jean-Michel Texier said in a statement. “Matching simple keywords against hundreds of thousands of documents sometimes returns a lot of irrelevant results. 3S provides an entirely different experience.”
3S can take in content from many different indices from many different Web publishing platforms and then apply semantic enrichment to it. The embedded Text Mining Engine identifies concepts, categories, proper names, places, organizations, sentiment and topics in particular content pieces and annotates those documents to create a semantic fingerprint that exposes underlying nuances and meaning, according to Nstein.
“Not only can we tweak the search sensitivity on-the-fly, we can also control content placement and even ad placement,” Gesca Digital Chief Technology Officer Matthieu Delorme said in a statement. “If we want to push a certain piece, 3S makes certain the right users find it.”
Nstein’s first 3S client, Gesca Digital expects to use of 3S capability to deploy topical microsites. “You input a search or a combination of searches across your properties, and channel the results into an RSS-ready template for a topical microsite, just like that,” Delorme added.
The unit of Montreal-based Gesca Ltd. — a Power Corp. of Canada subsidiary that publishes Quebec and Ontario newspapers, as well as magazines and books, supplied television programming and operates www.cyberpresse.ca — is customizing the tool, as Nstein intended. 3S’s visual interface allows administrators to tweak search sensitivity without modifying code. A slider-based interface modifies results between sets of sensitivities — e.g., more recent vs. more relevant.
3S comes bundled with front-end widgets that can point users to similar content, most recent content, or other identifying characteristic. Integrators using widgets and the templating engine can quickly build complex search-based mashups, across indices.