Internet Answer Engines Are on the Rise

Let’s answer your first question: What the heck is an answer engine?

An answer engine is a search site that provides you with an actual answer to a question, rather than a list of sites where you must find your own answer. If you were to do a search on “answer engines,” your result would be a list of several sites, such as the Wikipedia page for “answer engines,” a smattering of actual answer engine sites, and a few blog entries or news items on the topic.

The idea of a natural-language Q&A-based search engine is possibly as old as the idea of search engines. But what is fresh about this field -- and therefore inspiring tons of new startups -- is that people are in a bigger hurry than ever. For simple factual questions, searchers don’t always want a list of sites, they just want to know the answer. This is especially true when they’re searching from a mobile device.

There are a few ways answer engines go about providing answers: using pure computational know-how, hiring people to find the answers, and a combination of both methods. While the idea of people answering search questions can bring to mind the opinionated din of community-driven crowd-sourced sites like Yahoo! Answers (the granddaddy from 2005 that’s resolved 89 million questions) or Answerbag, many answer engines have expanded into more sophisticated territory by aiming to provide quick, definitive answers to factual questions or, for more esoteric questions, combining human answerers with the power of social networks.

Here is a smattering of both new and old players in the market. The services are divided mainly into social or nonsocial (or, one could say, opinionated versus factual) categories.

Nonsocial/Factual Answer Engines

Swingly: Brand new on the scene and still in invite-only beta, this small Texas-based company claims to already have more than 100 billion Q&As in its database, which it develops by using natural-language processing to turn Web page content into Q&As. According to the company’s blog, some of the results are from human-powered sites like other answer sites or Wikipedia, but the vast majority are created by its spidering technology.

Wolfram|Alpha: Also grabbing attention is Wolfram|Alpha, which bills itself as a computational knowledge engine. It claims not to be a competitor to Google search, but rather deals in a different type of knowledge: facts, statistics, charts, and data driven from a large database the company has gathered from public and private information. The company has a staff of more than 150 to work with that data, cleaning it up and tagging it properly. Wolfram|Alpha is more of an “invisible Web” search engine -- searching within data that usually isn’t crawled by search engines -- that also has the ability to do complex mathematical computations. This is the engine for heavy duty business research.

Ask.com: This old-school player recently launched a new, invite-only beta that combines its newly enhanced Q&A technology with human answerers from the Ask.com community for complex and subjective questions. Answers show up at the top of the page for easy identification. The site can push questions to specific people based on the content, and those responses will be indexed for the future. According to Ask.com, it has 500 million Q&A “pairs” indexed.

ChaCha: This answer engine for mobile users is human-powered. Independent agents receive a small (in the cents) fee per question they answer; the fee depends on the complexity of the query. The best responses are archived for quick answers in the future. The service is free but ad-supported for users.

KGB Answers: This effort, started by a 19-year-old international provider of directory assistance, offers answers via texts (#542542) or the Internet for 99 cents per question (monthly plans also exist). The answers are provided by human “agents,” and the company has expanded its services with kgbdeals.com, a coupon site with daily discounts. According to the company, KGB answered nearly a billion questions in 2009 and served more than 100 million consumers internationally.

Answers.com: Answers.com is a hybrid model, with two products that make the best of social crowd-sourcing and recognized sources. Its WikiAnswers is a community-based Q&A platform using wiki technology to create a giant Q&A database, while its ReferenceAnswers relies on licensed content from top reference publishers.

Social/Opinionated Answer Engines

Aardvark: Perhaps best known for being purchased by Google last year (and being founded by former Google employees), Aardvark taps into your social network. When you ask a question via chat, e-mail, or Twitter, it is farmed out to someone in your social network who Aardvark’s software has determined is an expert on the subject of your question. This can be a one-degree friend or a friend-of-a-friend. Next thing you know, you’re chatting with someone who is answering your question, solely for the sake of helping you out. It’s not just a good way to get a subjective question answered -- it could broaden your network, too.

Facebook Answers: Another big company getting into the answers game is Facebook. Facebook recently began rolling out a beta service to select users wherein someone can poll all of Facebook or just friends with a question. However, due to the lack of anonymity (questions and answers are tied to profiles) and general openness, this service may end up resembling something like Yahoo! Answers rather than a strict answer engine.

Quora: Run by ex-Facebook execs, this service has users log in through their existing social networks, but it also offers a way to answer questions anonymously. Questions are treated like threaded conversations, so it’s easy to keep an eye on what’s happening with a question. Like many other social answer sites, users can be endorsed and therefore rise up the ranks.

Mahalo Answers: It makes sense that Mahalo, the “human-powered search engine,” would also add a human-powered Q&A component. Mahalo Answers allows Mahalo users to tip each other for best answers, and some answers are incorporated onto search results pages.

When to Use an Answer Engine?

To take a human-answerer approach, it depends. If you have a very factual question (you want to know the date of a historic event, for example), a site like Swingly, ChaCha, Answers.com, Wolfram|Alpha, or KGB Answers is a good, quick bet. If it involves math or statistics, go with Wolfram|Alpha for now. And if you have a subjective question -- regarding anything from travel or food suggestions to public opinion about current controversies -- go with a social source. You’ll have to hunt through answers (though typically, the best-rated answers float to the top), but if you’re looking for an opinion, that’s exactly what you’ll want to do.