If you would like your company to be considered for an interview please email editor[ at ]semanticweb[ dot ]com.
In this segment of our “Innovation Spotlight” we spoke with Elliot Turner (@eturner303), the founder and CEO of AlchemyAPI.com. AlchemyAPI’s cloud-based platform processes around 2.5 billion requests per month. Elliot describes how their API helps companies with sentiment analysis, entity extraction, linked data, text mining, and keyword extraction.
Sean: Hi Elliot, thanks for joining us, how did AlchemyAPI get started?
Elliot: AlchemyAPI was founded in 2005 and in the past seven years has become one of the most widely used semantic analysis APIs, processing billions of transactions monthly for customers across dozens of countries.
I am the Founder and CEO and a serial entrepreneur who comes from the information security space. My previous company built and sold high-speed network security appliances. After it was acquired, I started AlchemyAPI to focus on the problem of understanding natural human language and written communications.
Sean: Can you describe how your API works? What does it allow your customers to accomplish?
Elliot: Customers submit content via a cloud-based API, and AlchemyAPI analyzes that information in real-time, transforming opaque blobs of text into structured data that can be used to drive a number of business functions. The service is capable of processing thousands of customer transactions every second, enabling our customers to perform large-scale text analysis and content analytics without significant capital investment.
Examples of customer use cases include: Analyzing tweets for positive / negative sentiment, processing SEC filings for companies to drive automated stock trading platforms, social media monitoring and business intelligence, powering recommendation engines and contextual advertising platforms, semantically-powered search engine optimization and competitive intelligence.
Sean: Can you give us a few examples of how companies are monetizing your API?
Elliot: Yes, we have quite a few well-known companies using our API for interesting applications:
* PR Newswire leverages AlchemyAPI to power its “Agility” social media monitoring and targeting platform.
* Salesforce.com’s Radian6 platform uses AlchemyAPI to perform large-scale crawling and analysis of news and blog content.
* Waggener Edstrom uses AlchemyAPI to identify key “Internet influencers” for specific topics and to help PR professionals target stories to the press.
* A number of Internet advertising networks use AlchemyAPI to provide better, more-targeted contextual ads, incorporating sentiment analysis and other advanced features not seen in competing keyword-based ad platforms.
* Business intelligence software providers use AlchemyAPI to understand their customer’s unstructured data, providing key insights into ongoing operations.
Sean: Where does your API draw its information from?
Elliot: AlchemyAPI operates in three modes:
1. It is capable of analyzing any content sent directly to the service via the API. Customers use this mode of operation to analyze internal data, document repositories, or other content archives.
2. It provides web crawling functionality, making it possible to analyze content that exists on the public Internet. Social media monitoring firms and other companies use this capability to analyze blogs, news websites, and other online content sources.
3. It can integrate with “Data Firehose” providers, such as GNIP or MoreOver, enabling real-time analysis of billions of tweets, social media updates, and other content.
Sean: How does your API work with Linked Data and what types of applications are people using this capability for?
Elliot: AlchemyAPI was one of the first semantic analysis platforms to embrace RDF and the Linked Data standards, and has provided these capabilities for a number of years.
AlchemyAPI links semantically analyze text to 3rd-party data-sets in the “Linked Data Cloud” such as: Dbpedia, Freebase, Opencyc, Geonames, and many others. AlchemyAPI customers can then access structured data that describes the people, companies, organizations, and topics mentioned in analyzed documents.
We’ve seen a number of interesting usages of Linked Data. For instance, an academic research group constructed a “Disease Tracker” application that tracked the outbreak of diseases across different cities, states, and countries. They leveraged Linked Data to provide real-time demographics, life-expectancy stats, and other information for regions experiencing disease outbreaks.
Sean: Your API offers a lot of features. Which ones are most popular?
Elliot: We offer a bunch of APIs (12+) for analyzing content in a variety of ways. Named entity extraction, sentiment analysis, keyword extraction, and fact/relation extraction are some of our most popular APIs. However all of our APIs incur significant usage within any 24-hour period.
What techniques are you using for Entity extraction?
AlchemyAPI uses a hybrid approach incorporating deep linguistic processing and statistical analysis. We’re tokenizing text, breaking it down into its basic components (nouns, verbs, etc.), applying shallow and deep linguistic parsing techniques, co-reference resolution, and finally a process we call Named Entity Disambiguation.” This linguistic approach is supplemented with a statistical view of the data. We’ve found a hybrid approach provides the best accuracy across a wide variety of content sources: From grammatically-correct editorialized text all the way down to badly written misspelled tweets.
Disambiguation enables us to go a step further than many other entity extraction solutions, answering questions like “Is this Paris TX? Or Paris France? Or Paris Hilton?”
Sean: How does your sentiment analysis API work? Are you able to analyze raw data from social networks like Twitter and Facebook?
Elliot: AlchemyAPI has 5 different sentiment analysis APIs, providing a wide range of capabilities for looking at opinion-based content. This includes document or sentence-level APIs, entity and keyword-targeted APIs, user-targetable APIs, quotation-based sentiment analysis, and even “directional sentiment analysis.”
Directional sentiment is interesting in that we expose not only who/what the sentiment is expressed towards (positive sentiment towards Coca-Cola), but also who is responsible for that opinion. This enables you to determine not only the “why” but the “who.”
Our service processes many millions of posts from social networks such as Twitter and Facebook every day for our customers and partners.
You also have a Keyword Extraction feature. Can you tell us how this works?
Keyword extraction is a complimentary API to named entity extraction, focusing more on identifying abstract concepts such as green energy, quantum physics, etc. Named entities, on the other hand, provide information on real-world people, places, companies, and things.
AlchemyAPI uses a hybrid approach here as well, incorporating deep linguistic parsing and statistical analysis, to identify topical keywords, ranking, scoring and computing their sentiment. Our ad network customers love this API.
Sean: Do you have a free trial for users to test your API?
Elliot: Definitely, you can test the API at: http://www.alchemyapi.com/api/register.html. It allows developers to sign up and process up to 1k API calls per day.
Thanks so much for your time Elliot!
Join the discussion. Please let us know what you would have asked below.
About the Author:
Sean Golliher (@seangolliher) is an adjunct professor of search engines and social networks at MSU and is a member of their computer science advisory board. He is also the founder and publisher of SEMJ.org. Sean holds four engineering patents, has a B.S. in physics from the University of Washington in Seattle, and a master’s in electrical engineering from Washington State University. He is also president and director of search marketing at Future Farm, Inc., Bozeman MT, where he focuses on search marketing, internet research, and consults for large companies. He has appeared and been interviewed on well-known blogs and radio stations such as Clickz.com, Webmasterradio.com, and SEM Synergy. To maintain a competitive edge he reads search patents, papers, and attends search marketing conferences on a regular basis.
- Bringing Startups Center Stage at SemTechBiz 2013
- GraphLab Raises $6.75M to Build 'Hadoop for Graphs'
- The Latest Breakthroughs of E-Discovery
- Semantic Web Company and OpenLink Partner to Advance Enterprise Linked Data Integration