EVRI, the semantic content discovery vendor, has talked about focusing its attentions on two things: Helping consumers make sense out of growing volumes of information, and exposing more of its functionality to developers via APIs so they can create their own takes on solving that problem.

Last week the company announced its sentiment web API, which lets developers build applications around the sentiments of specific entities as well as categories. The company expects early adopters of this API to include those working in areas such as market intelligence, says Deep Dhillon, Evri's CTO, in an email interview.

Other potential applications that may have more direct applicability to the consumer include helping them ferret specific sentiment-related information around product reviews or tracking commentary in the worlds of sports or entertainment. The sentiment API can be used to inform applications to facilitate the discovery of what percentage of things being written about the iPhone are positive or negative expressions of sentiment; who is criticizing or praising a particular thing or person or place; or who a particular known entity -- whether celebrity or organization -- is praising or complaining about, among other things, according to Dhillon's posting in the EVRI blog.

The sentiment API, Dhillon believes, is another step towards letting developers help solve consumers' problems around the web content explosion.

"Our API is all about exposing the ability to, in a fully automated way, make sense of high volumes of data," he tells Semanticweb.com, noting that for one customer EVRI has performed deep semantic processing for over 90 documents per second. "We are continuously innovating on new ways to expose more distilled information about the web of entities being continuously written about in the news, blogosphere and other web content. For example, our newly launched sentiment API lets developers have access to how these entities feel about each other, which people, places or things they are criticizing, and which they are praising."

Market Intelligence Needs

The sentiment web API aspect joins the media recommendation functionality within EVRI's API that is being used primarily by publishers, enabling developers to, given an article, create a recommendation experience that allows users to discover fresh news articles or blog posts to read, related images to view, or full-length related videos to watch. It also accompanies another aspect of its API, which has to do with identifying entities from a given piece of text and linking this identification to EVRI's knowledgebase of entity information.

EVRI says it pays close attention to delivering the APIs the development community wants; Dhillon notes that one of the drivers behind the sentiment web API was related to the needs of those in the market intelligence area.

"They really want to not just track the mentions or comment/tweet velocity of a product or company, but they want to go much further and know how it is being talked about on the web -- that is to say, whether people are praising it, or bashing it," he says. "I suspect folks wanted us to tackle the problem, in part because they don't want to have to use yet another API and be dependent on another vendor."

In addition, he thinks EVRI's deep technology stack which lets us go further than other offerings: "For example, most sentiment analysis packages will simply state a percentage of positive or negative sentiment, and then you can view expressions in one category or the other. One thing we do which is really different is we actually say who or what is expressing the positive or negative sentiment, as opposed to simply remaining ambiguous," he says. "These more involved capabilities are possible because of our semantic depth -- because, for example, we know for every sentence and grammatical clause what the subject, verb, object, modifier information, prepositional information, etc. is, we are well poised to reason behind simple membership in a category."