Today we rolled out a new release on Center’d that introduces “Flavored Local Technology” – a machine learning based approach to classify local places of interest based on conversations on the web from locals. We have been providing the collaboration and planning tools for almost an year now and the addition of this technology and the related tools should help you plan from a casual plan (where to go dinner tonight with kids) to an elaborate plan (how to organize the summer picnic at school). And that is exactly what we would love to do – help you plan your life’s activities.
Intent based activity search is one of the best ways to approach casual local planning problem – it is a new kind of local search with activities or things to do being the primary focus; and if you think about it activities are driven by the intent – when you look for a restaurant, you have a specific intent (say romantic?); when you look for an attraction you have a specific intent (say group friendly?); when you search for things to do (say kid friendly?), you have an itent; today with this release, we are launching five (5) supported intent “styles”: kid friendly, romantic, group friendly, outdoor and “recession buster” (aka cheap) to make it easy to find places that fit into popular “categories” such as eat (restaurants etc), visit (attractions), attend (local events) and shop (shopping).
Crawling the web for local conversations and understanding the intent/activity behind them is a huge task – eventhough we used the cloud-computing goodness with Amazon EC2 compute clusters – there is simply not enough time to get everything done right – so we are rolling out this release in the following top 12 cities: Atlanta, Boston, Chicago, Houston, Las Vegas, Los Angeles, New York, Palo Alto, San Diego, San Francisco, Seattle and Washington DC. Other cities will follow soon. Of course, we still have the local search enabled for other cities, but we just haven’t finished processing the data yet. So, give it a try and give us feedback!
I’m excited the way this has come together – incredibly hard-working team (can you believe all the development was done by just 4 engineers?) coupled with my learnings from IR tinkering put to work to solve a good problem: enabling global access to local knowledge. Now you be the judge – give us a try and let us know!
Here is the copy of the full press release:
Center’d Unveils “Flavored Local” Search – New Approach to Local Planning
Filters Content by Intent
Develops Unique Activity-Based Search Index of More than 1 Million Places
Intent-Based Styles Including “Kid-Friendly” and “Recession Buster” Help People Quickly Find Things to Do That Meet Their Needs
APRIL 28, 2008 – MENLO PARK, CA – Center’d (http://www.centerd.com), the Web site that helps people plan life’s activities, today released new tools, features and data that help people find things to do in local areas based on their intent. By analyzing millions of conversations about places on the Web, Center’d has created a unique index of more than 1 million places that are classified by activity intent, including kid-friendly, romantic, cheap, and more. This allows people to quickly discover places and activities that best suit their needs while short-cutting the overload of unstructured local data on the Internet. To highlight the new personal planning information and tools, Center’d has redesigned and enhanced many areas of its site.
“Consumers want a faster way to find local activities that meet their needs without having to visit multiple sites and sift through hundreds of reviews. Our unique activity index and personal planning features help solve this problem and complement our existing group collaboration tools,” said Jennifer Dulski, co-founder and chief executive officer of Center’d. “In addition to solving a consumer pain point, this new approach to local search also positions us well to provide advertisers and publishers with more relevant solutions to engage audiences and monetize content.”
A Novel Approach to Local Search
Center’d created its unique activity-based index using innovative natural language processing (NLP) and machine learning technology to analyze millions of conversations across the Web. The index includes inputs from local review sites, aggregators, and blogs through feeds and publicly available sources, as well as data from plans people make using group collaboration tools on Center’d. Based on its data analysis, Center’d has integrated 5 intent classification “styles” on its site which include kid-friendly, group-friendly, romantic, cheap, and outdoor. As data continues to be analyzed and classified, more intent styles will be made available on the site.
People can currently find information about restaurants, attractions, shopping and events across the intent styles in a number of new or enhanced areas of the site:
Enhanced Search: Center’d has built upon its ability to filter search results by people’s social graph and added filters for intent styles, which yields more accurate results for activity-based searches compared to standard keyword search.
New City Pages: Center’d has created city guides for 12 major cities where people can browse things to do by intent style and place category. This allows consumers to get a view of a city filtered by their personal needs. For example, each city guide will have a kid-friendly style page with restaurants, attractions, events and movies that are suitable for children. The new city pages also feature editorial content from bloggers with local expertise that complements the intent styles.
Updated Place Profiles: To give people a faster understanding of a business or attraction, place profile pages now include the most popular “snippets” of web conversations, organized by themes such as service and ambiance. These snippets are coupled with sentiment graphs that show what percent of comments are positive, neutral or negative. Together these tools give people a quick way to determine the tone of Web conversations about a place.
Redesigned Home Page: The new Center’d home page highlights enhanced functionality which includes activity styles and personal planning features, as well as existing group collaboration tools. People can quickly pick a style (e.g., romantic) and type of activity (e.g., restaurants) that suit their needs, and Center’d will present relevant search results.
“Distilling the unstructured data available on the web about local places and events is a mammoth task. The technology we’ve built to address this provides a foundation for us to deliver unique mobile and Web applications to consumers and partners going forward,” said Chandu Thota, co-founder and chief technology officer of Center’d. “The investment we have made in developing this unique index will prove an increasingly unique competitive advantage as we continue to scale our business.”
New Data Helps People Save Money
As part of the 5 intent classification styles announced today, Center’d has included a “cheap” category designed to help people save money while still enjoying local activities during the recession. Based on people’s Web conversations Center’d has identified places that are inexpensive, have special promotions or that are free in general. Places listed in this category include things like cheap restaurants, museums with free admission days, parks and free events, among other places and activities. People can find cheap things to do in their local area by conducting a search from the new home page, browsing a city guide page, or by filtering search results.
In conjunction with this product release, Center’d has created partner APIs which are currently available by contacting the company directly at bd(at)centerd(dot)com.
Center’d (http://www.centerd.com) helps people plan life’s activities. The company has developed personal planning features and group collaboration tools that help people plan any type of activity, from finding and discovering things to do, to organizing and coordinating complex events. Using natural language processing and machine learning, Center’d has created a proprietary activity index that allows people to find things to do based on their intent. This unique approach and its proprietary activity index enable Center’d to deliver a more relevant and personalized local experience.
Headquartered in Menlo Park, California, Center’d is led by former Microsoft, Yahoo! and Amazon executives, and is funded by Norwest Venture Partners and KeyNote Ventures.