Ron Miller
SemanticWeb.com Contributor

Inbenta, a Barcelona-based software company that develops software which can understand natural human questions, has a free version you can add to your small web site or non-profit site, while offering other customers a pay version of the software.

Who is Inbenta?

Inbenta was started back in 2005 by a unique team of computer and social scientists, according to Jordi Torras, director general at Inbenta. “A combined team of IT developers and computational linguists work together to make our software understand human questions and give relevant answers,” he says. Inbenta technology attempts to understand the intent of every question, the actual meaning of every search query beyond what the user actually typed in. “Natural Language processing allows us to identify that queries like ‘price for your products,’ ‘how much is it,’ and ‘what’s the cost’ are almost equivalent in meaning, and therefore have similar search results” Torras says.

This is in stark contrast to a keyword-based search like Google or Bing, which he explains is more random and prone to user error. “Most users are used to trying different wordings and phrasing for their search queries until they find what they actually wanted–that ends up annoying users and therefore leading to lost sales and needless calls to contact centers.”

How it Works?

Torras says that this technology is not meant to be used on the open web, but for externally facing web sites to help visitors find the answers to their questions. “Our software could be described as a ‘meaning-based, enterprise-wide search engine,’ as it compares meaning of user queries with the meaning of contents, finding those items that have a reasonable similarity in meaning only,” Torras says. The software is designed to help users find answers without having to call the company. “The aim of our software is to have a much higher relevance at website search applications, and thus increase online conversion and decrease customer support costs.”

He gives an example you can try on the Inbenta website, which itself (as you might expect) uses this technology. “If you ask at Inbenta’s website ‘How much is it?’ you will get a result with this title: ‘How can you ask for an estimate?’” He explains that asking the same question at Google.com while restricting results to inbenta.com, (using the following search query: ‘How much is it site:inbenta.com’) you will get irrelevant answers, such as ‘Inbenta in the Catalan Computational Processing Meeting- Mar 2009.’ “Google does that because the word ‘much’ exists in that particular document, but the words ‘how,’ ‘is’ and ‘it’ and the [underlying] meaning are completely ignored,” he says. Torras says that this is true for all keyword-based search engines.

Where are the Semantics?

If you’re looking for semantic underpinnings like OWL, RDF and RIF, you’re not going to find it here, but he says, “Our search algorithm is based on a concept call “Lexical Functions” which is, in turn, part of the “Meaning-Text Theory” [developed] by researcher and doctor Igor Mel’čuk.” On top of the base search algorithm, Inbenta has a ‘linguistic corpus,’ a database containing hundreds of thousands of words and their semantic relationships, expressed as lexical functions. Torras says, this linguistic corpus is available now for all major European languages, and is maintained by linguists at Inbenta, where it is continuously evolving and growing.

Mills Davis, managing director at Project 10x, a consulting firm specializing in semantic technologies, says that natural language processing does in fact fall within the spectrum of semantic technologies as he sees it. “A key requirement (I believe) is that it has to be able to read, semantically annotate, and index whatever it is that you have in or attached to your site,” he says, and natural language processing meets this criteria.

Davis adds that its semantic underpinnings are not necessarily as important as how useful it is to the end user. “Actually, the proof of the pudding will be the user experience. If it’s ‘semantic’ in a useful way, then they should be able to surface some very helpful features for people searching, querying, and navigating the site.”

Inbenta may not be part of the semantic web as we have come to understand it, but it is part of a growing number of enterprise products taking advantage of semantic technologies. Natural language processing like that offered by Inbenta can help enterprise customers achieve their goals of helping visitors find answers online, today.

The free version of Inbenta’s software has some limits. “It is free for up to 500 pages to index and less than 100,000 search queries per year,” Torras says.

Semantic Tech & Business Conference Returns to San Francisco

Semantic Tech & Business Conference returns to San Francisco in June! Join us from June 3-7 for complete coverage of Big Data, Linked Data, Extreme Information Management, and Semantic Web. From breakthrough approaches to solving business problems to the big data implications of fast–evolving technologies, SemTechBiz provides you with an unparalleled interactive experience and delivers tangible business value. We're offering a special early rate when you register by February 17. Sign up now!